The GetDPI Photography Forum
Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!
Sigma DP2 Quattro Shots
- Thread starter Quentin_Bargate
- Start date
Quentin_Bargate
Well-known member
One has to admire the enthusiasm of Sigma CEO Kazuto Yamaki, but is he too close to the product to see its faults?
Kazuto Yamaki's Presentation at CP+ 2015: Sigma dp3 Quattro Due Out in March, Technical Tidbits
An English translation would be helpful.
Kazuto Yamaki's Presentation at CP+ 2015: Sigma dp3 Quattro Due Out in March, Technical Tidbits
An English translation would be helpful.
Hulyss Bowman
Active member
Thank you for coming today, and thank you for supporting our product and activities.
I'm giving two presentations about our new product at this CP+ 2015, and today, I'd like to talk about our new cameras.
Here is a box but there will be no surprising items. I'm not so resistant to the pressure, so I'd like to say that I'm talking only about previously announced new items.
We announced the new dp Quattro series, dp0. After the announcement, I've got lots of comments that say “it's crazy!” Yesterday, other company's staff talked to me saying “SIGMA is trying to maximize it's craziness, isn't it?” I said “No, it's not our intention.” I feel some weird atmosphere.
This is dp0. I'm going to talk about this in detail later. It's been a long time since we started selling the first DP1 that a lot of customers have wanted to buy a little wider version of DP1. Even I often feel the same way when I use DP1 personally. As DP1 can capture the detail, I encountered various cases that I feel if only I have a wider lens DP. However, we couldn't decide how wide it is possible as business.
First, we started making a wide converter lens to achieve the 21mm focal length. But it was difficult because the image quality of the corners was blurred. I know that a wide converter lens is convenient, so I don't mean to say that it is useless. However, thinking about the concept of the DP, it might not be good to sell a wide converter lens at first. So we abandoned the project.
Whenever I travel around the world for business, I have been asked about the wider DP for a long time. So we finally started the project officially last year to make the wider DP with a special lens.
Today, I'd like to introduce this dp0 and dp3 which will be available soon. But before that, I want to talk about the concept of the dp Quattro series. This might share the same information that I talked last year, but please let me explain the idea.
The goal of the dp Quattro is to achieve the image quality with a tangible and realistic feel. We do not compete to increase the pixel number, but to acquire the higher image quality. My hope is to achieve the image quality which the medium sized film cameras used to have.
This is an off-topic episode, but Foveon's chief engineer likes to shoot with medium size film camera along with developing the image sensors. He is a unique guy. But unfortunately he is so busy lately that he has less chances to go out shooting. Anyway, to reach the goal, we SIGMA team and Foveon team collaborate each other with the same direction.
So how to acquire the “rich” image quality? We think it is necessary to get minute detail and stable detail for any kinds of objects. For DP camera, we think three factors are important. The first is the Foveon X3 image sensor. The second is the specially designed high quality lens. The third is the fixed, not interchangeable, lens which uses a lens shutter with less vibration and is aligned perfectly to the image sensor. An interchangeable lens camera relies its image quality on the accuracy of the lens mount as they align the image sensor to the camera mount and align each lens to the lens mount. On the other hand, this fixed lens camera is designed to adjust the matching of the lens and the image sensor from the center to the corners. So, the lens is physically “attached” to the image sensor to achieve the perfect resolution anywhere on the image. This is the characteristic of the lens fixed camera.
You might hear this many times, but please let me explain the Foveon X3 image sensor.
This is a normal Bayer sensor, which can not distinguish the color difference so it uses the RGB color filter to obtain color information at each pixel. Then, it compliment color information by calculation and make a full color image. This Bayer sensor is named after its inventor Mr. Bayer. We Japanese call him “Bay-yah,” but actually the pronunciation is “Buy-yah.” I heard his bereaved family also want to be called “Buy-yah,” anyway.
The next is the Foveon sensor. This sensor can capture different wave lengths depending on the depth of the silicon. Here is also what I talked last year. For normal color filter sensor, the green shares 50% of the pixels and the blue and the red share 25% each. On the contrary, the Foveon sensor can capture all RGB at any pixel. The reason why normal sensor spares 50% of the pixels for green is that human eyes are the most sensitive to green. The shorter wave length is blue, the middle is green, and the longer is red. Longer than red is infrared, and shorter than blue is ultra violet. Both human eyes can't see. As Foveon sensor can captures green at all pixels, if the number of the pixel is the same, Foveon has twice as high resolution as Bayer sensor. This is the principle of the Foveon sensor.
I'm surprised that a lot of audience here has Quattro cameras, as I have few chance to see someone with a Quattro camera on the road. Thank you so much.
The characteristics of the Quattro sensor is that on the top layer, it captures the detail of the object. It consists of 20 mega pixels. For colors, the top, the middle, and the bottom layers captures each color. The middle and the bottom layers are four times bigger than the top layer. As Foveon is three layer sensor, the data size is triple, and because of this, the processing, transmitting, and writing the data get slow. However, Quattro can solve this problem. Anther difficulty is that the wiring becomes complicated if we maintain the previous structure. The size of the photo diode becomes smaller and the amount of intake of light also becomes worse. By using the Quattro 1:1:4 structure, we do not sacrifice the intake, and could maintain the detail of Foveon X3.
These are the specifications of the Quattro sensor, which you might be aware of. The total number of the photo diodes is 29,400,000. It is comparable to the 39,000,000 pixels Bayer sensor in terms of the resolution.
With the Foveon sensor, whatever the object is, you could get the detailed image. I think the uniqueness of the photography is its reality, compared to other arts such as paintings. We would like to pursue the expression which has reality as if it exists in front of you. Therefore, it is necessary to avoid unnatural resolution, such as some area is detailed but other is not.
Here is a resolution chart taken by dp2 Quattro and 36 megapixel Bayer sensor. Both sensors' resolution is quite good, but on the Bayer sensor, the brightness resolution and the color resolution don't match. Therefore, there is false color. On the other hand, even it's structure is unique 1:1:4, Quattro's final image has no false color, that is, you can obtain a clear and clean image. This is the characteristic of Foveon, and with this image, we could achieve the “middle size quality.” These are the technological background of Foveon.
Before I started preparing this presentation, I asked some questions on twitter. Someone asked me “Is there any room to improve the quality of the Quattro sensor?” He has been using SIGMA camera for a long time, so he knows a lot. I felt it's a really critical question.
I have something to apologize for users. Sigma Photo Pro ver. 6.0 was released with dp2 Quattro, which is designed to develop Quattro file. Current version is 6.2. But previously, there are some errors and unoptimized processes on the SPP. For example, if you take a long exposure photo, the noise reduction works too strongly, so the detail had disappeared. I knew a lot of users wanted to take a detailed long exposure, but there was a bug on the SPP.
The second is that there was too much noise reduction on the shadow area. As I said, we want to make a high resolution camera with all kinds of objects. But there was a problem, though it's been fixed already. Finally, as we developed the new 1:1:4 structured sensor, which is totally different from its predecessors, we didn't understand everything and found there were ways to improve the image quality.
We have updated and keep updating the firmware to optimize the Quattro sensor. The next version SPP 6.2.1 is coming soon. We are planning to release this at the end of February. If you have a Quattro file, please download the new version, and re-develop it. I think a lot of users were worried about the Quattro image, but I'm confident that SPP 6.2.1 will satisfy you. We will keep working to improve the Quattro image and developing the appropriate algorithm.
Another question is “Can SPP processing become faster?” and “Are you going to sell a high speed PC for Quattro development?” Of course, we are not planning to sell a PC, but we understand that the slow processing speed is a problem. We have improved the speed quite a lot at the SPP version 6.2. However, it is also true that we have prioritized the processing quality rather than the speed. So we will focus more on the processing speed and put much effort to improve the usability of SPP.
I'm going to talk about the new product. The first is dp3 Quattro. As you know, it has 50mm lens, which is equivalent to 75mm on a full-frame. dp1 is equivalent to 28mm, and dp2 is equivalent to 45mm. In the series, dp3 is positioned as a middle telephoto lens camera.
Now, we have a new ultra-wide lens dp0. To say the truth, I'm a little worried about the naming of dp0. Someone wrote on twitter that if there will be a wider lens, will it be dp minus 1? The other said if there were 35mm dp, would it be dp 1.5? Even SIGMA couldn't release something 1.5, so I was wondering the name of dp0. But finally, as I'll talk about this later, the distortion of dp0 is almost zero, so it's wider than dp1, and zero distortion, dp0 represents these ideas.
Anyway, let's talk about dp3. As some of the DP3 users might notice, it has a really sharp lens. It can give a real DP quality image. The design of the lens is really classical. It consists of a double-Gauss master lens with a converter. Because its focal length is standard, we can choose a classic lens design, but we put the cutting-edge glass element into the classic design, and manufacture it at the most sophisticated factory. So it achieves the ideal performance of its design, and the yield rate is also high. A lot of users appreciate its image quality, saying it can take sharper image. Compared to all SIGMA lenses, this dp3 lens is one of the sharpest lens.
Updating to Quattro, we improved the AF speed. It's 20 to 30% faster than the previous DP3 Merrill. dp camera has a problem on focus at low light, but we also improved this issue. In addition, dp2 and dp1 Quattro's AF will improve with a new firmware, which will be released at the end of February. The exact day has not been decided yet though. So for Quattro users, please download the new firmware and try it.
The release of dp3 Quattro is in March. So, please think about it.
We also announced the 1.2 times conversion lens FT-1201 for dp3 Quattro. The equivalent focal length will become 90mm, and as it's a front converter, the F-stop is the same 2.8. The concept of this is the converter that doesn't deteriorate the image quality of the dp3 master lens.
This is the picture of the converter. As you can touch there, it's a big converter consisting of three groups and four elements. Here is the MTF chart. The left side is the dp3's master lens, and the right is the dp3 with the converter. The left side of the chart shows the center of the lens, and the right side is the corner. The red line represents low frequency, such as a thick big object, and the green line represents high frequency, small objects. From the center to the corners, both lines are flat and the performance is high.
With the converter, the performance at the center is the same. From the center to the corner, the performance is stable. But please check the far right. At the corner, the performance is actually higher than the master lens. I was wondering whether it is true, so asked to the designer. He said it's a coincidence. What a nice coincidence to get a better performance. I wouldn't say that all performance is better than the master lens, but the MTF chart is like this.
The picture of dp3 with the converter is here. The original lens consists of 10 elements, and the converter has four elements, which means 14 elements. As the body of the dp is thin, it becomes a front heavy camera. You might be worried about the balance, but please feel at ease, we have a LCD view finder. With this finder, the balance of the camera gets better.
Let me talk about the finder a little. The part which attaches to the camera body is called bracket. This part is made by aluminum machining, so it's really solid. The eyepiece lens is optical glass, which is as same as an interchangeable lens. The diopter adjustment uses an advance helicoid, so its feel is as sticky as a manual focus lens.
Here is dp3 with the converter and the view finder. As this becomes 90mm, it is better to attach the eyepiece to your left eye to avoid the camera shake. It will be a good portrait camera. But to say the truth, I can't understand what I'm doing on this camera, seeing its form. Anyway, please try using them.The converter is going to be released with dp3 next March, but I'm not sure whether it's the same day.
We announced the bottom grip too. We try matching the appearance to dp cameras. The top part is processed with alumite treatment which is the same process we used for dp's manual focus part. The knurling pattern is also as same as the dp's camera dial, so it matches really well. The screw is made of stainless steel, not aluminum, so it's durable.
Here is the sample. This is also made by aluminum machining, so you feel some heaviness and quality. If you attach it to the camera, it looks like this. Here is an image. Please try it. It will be available in April.
Then, let me talk about dp0. It has a really long lens, so when I check the customers' reaction on the net, they are saying “Isn't it 85mm?” Someone I met yesterday said “Are you putting a 10 times zoom lens into it, aren't you?” But to be honest, this is just an ultra wide angle lens. I'll explain the detail later.
The concept of dp0 is “zero distortion super wide camera.” Someone asked me on twitter saying “Why the lens is so long?” The reason is to achieve the dp quality. We try making the lens which could capture the minute detail of the object. The distortion is almost zero, and we corrected the chromatic aberration thoroughly.
We also want to put the light vertically into the sensor, so we achieved the higher telecentric performance. Because of these, the length of the lens became that long.
Here is lens composition. Actually, it is really fancy composition. We incorporated four FLD glasses which is resemble to fluorite and two SLDs, super low dispersion glass. The first large glass has aspherical surfaces on both sides. The final lens is also aspherical, so of all elements, normal lens is only three, and the rest is special glass, such as low dispersion or aspherical. So the composition is really gorgeous. Before I started preparing for this presentation, I talked with a designer of this lens, asking his intention and purpose. I said “Wow, it's really gorgeous and rich design,” and he said “Yeah, I've never expected that you would approve this composition.” I was surprised why he was saying at that time. Anyway, I'm happy that we could achieve high performance with this lens.
Here is a chart that explains the difference between normal glass and special glass. Through normal glass, the image blurs depending on the wave lengths. So there is an aberration. On the other hand, with special low dispersion glass, such as SLD or FLD, there is less dispersion, so less aberration.
By using these special glass a lot, the distortion of the final image is less than 0.5%. This is the least distortion of all dp cameras. Also, the chromatic aberration at the farthest corner of the image is one pixel. So the performance is really good.
The length of the lens contribute to achieve this high quality. As this is an ultra wide lens, you have to intake the light from a wider angle, and the light has to reach a large image sensor. If you bend the light drastically, it is difficult to avoid the appearance of the aberration. Therefore, we try bending the light gradually and carefully. Like an “In box daughter” (We Japanese call an only daughter brought up carefully like this), we avoid the light becoming delinquent, and carefully bend the light element by element. If you bend the light suddenly, there will be an aberration, so we have to correct it too, by sending her to a correctional institution. That's why the length gets longer like a telephoto lens.
Someone might wonder why we are working so hard to increase the lens performance. A lens distortion and a chromatic aberration can be fixed by the camera because it has a prime lens. However, this is also the same idea of the dp quality: to get clean and fine details. It's easy to understand the correction of the distortion. As the corners are smaller than the center, you have to enlarge these corners, then it will lose the detail. Although the software correction is more powerful and excellent, the restoration of the detail is impossible. We have to rely on the lens and the image sensor to get the detail. If you fix the distortion, you will lose the detail.
If it is a Bayer sensor, the correction is not distinguishable. On the contrary, as Foveon can get one pixel image, you could notice corrected area easily.
If I say so, you would think that it's not so difficult to correct the chromatic aberration as it is only the diffusion of color. Of course you could fix the aberration depending on the wave lengths, but it is not true.
Here is a simulation of dp3 and our classic 14mm F2.8 lens. The performance is terrible, but it's already discontinued, so it might be OK to show here. You can see the diffusion of color.
This is an excerpt of green light; the middle wave length. As you can see, even it's the middle wave length, the light is diffused. You can only correct the color with the chromatic aberration correction, and still you can't get the detail. Even the dp0 has an aberration, but no worries. It's an enlarged image, so the aberration is really small. Anyway, if you say you don't need to be worried about the chromatic aberration because you could fix it later, it's not true. With a less aberrated lens, you could acquire a sharp, fine detailed image. The goal of dp0 is this, so it has a long, large lens.
Here is dp0. It has such an extremely large lens that no one can believe it is a wide angle lens camera.
We also made a view finder of 21mm equivalent. I have no idea how many people would buy this, but anyway, we made it because it's necessary. It has a high magnification lens, and the eye point is relatively long, so you could see well with a naked eye or with glasses. The body is aluminum machining, and glasses are optical lens, so the quality is excellent. If this sells well, we could pursue the next project easily, so please think about it.
I haven't decided the release date, but I would like to start selling dp0 at the beginning of summer. With dp0, the dp Quattro series consist of cameras with ultra-wide angle to middle telephoto lens. This is dp0: 21mm equivalent, dp1: 28mm, dp2: 45mm, dp3: 75mm, dp3 with a converter: 90mm. Actually, it's very different, so you could call this dp4. As a whole system, they cover nice focal lengths. If you bring everything at once, it's a little bulky, anyway. But I believe we could make a good system.
Let me announce here about firmware update for dp1 and dp2 Quattro. It is scheduled at the end of February. As I mentioned, the AF at low light will improve. We add new aspect ratio, 7 : 6, new image size “super low,” a very small size. You could minimize the time lapse timing to 4 seconds. You could add and check the copyright information. You could use translucent frame when you change the aspect ratio. With this function, you could see an object outside of the frame. This translucent frame might remind some of you our SD9, which was also a “crazy” camera.
Someone asked me on twitter whether there will be a successor of SD1. Sorry, there is no surprise today, so I couldn't show anything. But we are working on the SD. Many people ask me about mirror less dp, but for us, the successor of the SD is a top priority. Someone named it freely like SD1 Quattro, but the name, the spec, nothing has been decided yet. We are working hard, so please give us a time.
We are often asked about Full Frame Foveon, but basically it's no comment. If you ask me if there will be a FF Foveon next month, I'd say no. But it doesn't mean there will be no FF Foveon in the future. I can't say yes or no.
What I want to say is that there are a lot of challenges when we make FF Foveon. Even with Quattro 1:1:4 structure, the data size will become huge. The data processing time and writing time will be a problem. More difficult problem is lens. The bigger the image circle is, the more difficult to design the optics.
At present, it is true that the APS-C size Foveon is well-balanced. As you saw, the normal Bayer sensor can get only half resolution as Foveon can. So for Bayer sensor, there is only one solution to increase the resolution: increasing the sensor size. Because there is a threshold to minimize the pixel pitch.
I believe that with a smaller sensor, Foveon is the only way to achieve the higher resolution, and APS-C Foveon can be comparable to FF Bayer sensor in its performance. If we make FF Foveon, considering the data size and other things, it will behave something like middle sized digital back. Please keep in mind these things. But basically for FF Foveon, it's no comment.
Finally, let me talk about the Foveon X3 technology. We are aware of its weakness, such as high ISO performance. As only we and Foveon are developing the sensor, it is difficult to increase the speed. But still, under some circumstance, Foveon can take a unique, special image. I believe it's a Foveon's raison d'etre. To respond to the customers' demand, we'd like to keep improving the sensor.
Another reason why we are working on Foveon is that I believe it has some historical significance. Some of you might know that the first color photography was French Lumiere brothers' autochrome. It was invented in 1903. They are also famous as the founder of the movie. These are the autochrome pictures taken at the beginning of the 20 century. The Lumiere brothers were raised in Lyon, France. I had a chance to go to Lyon last year, and spared some time to visit the Lumiere museum which exists in the suburbs. There were many pictures taken at the beginning of the 20 century, so I was really impressed. There were also some pictures taken in Japan during that time. It was really interesting.
So how they could make an autochrome image. They dyed potato starch into red, green, and blue, and put it onto the glass board. Then they exposed the glass. You might guess what I'm going to talk next, but anyway. Later, Kodak invented a color film, which captures colors vertically. So the color film era had started since then. The normal image sensor captures colors horizontally, though its performance has been improved significantly. However, the basic concept of the Bayer sensor is as same as the Lumiere brothers' autochrome. On the contrary, Foveon takes the “vertical” route which Kodak had broken through. We believe that this vertical approach has technological future, so we pursue this route. We have lots of challenges, but are working hard with hope.
We have a lot of ideas to implement into the Foveon sensor. It takes time to make everything feasible, but we SIGMA and Foveon are working really hard to improve the Foveon sensor. Please expect our future products and support us.
Thank you for listening.
I'm giving two presentations about our new product at this CP+ 2015, and today, I'd like to talk about our new cameras.
Here is a box but there will be no surprising items. I'm not so resistant to the pressure, so I'd like to say that I'm talking only about previously announced new items.
We announced the new dp Quattro series, dp0. After the announcement, I've got lots of comments that say “it's crazy!” Yesterday, other company's staff talked to me saying “SIGMA is trying to maximize it's craziness, isn't it?” I said “No, it's not our intention.” I feel some weird atmosphere.
This is dp0. I'm going to talk about this in detail later. It's been a long time since we started selling the first DP1 that a lot of customers have wanted to buy a little wider version of DP1. Even I often feel the same way when I use DP1 personally. As DP1 can capture the detail, I encountered various cases that I feel if only I have a wider lens DP. However, we couldn't decide how wide it is possible as business.
First, we started making a wide converter lens to achieve the 21mm focal length. But it was difficult because the image quality of the corners was blurred. I know that a wide converter lens is convenient, so I don't mean to say that it is useless. However, thinking about the concept of the DP, it might not be good to sell a wide converter lens at first. So we abandoned the project.
Whenever I travel around the world for business, I have been asked about the wider DP for a long time. So we finally started the project officially last year to make the wider DP with a special lens.
Today, I'd like to introduce this dp0 and dp3 which will be available soon. But before that, I want to talk about the concept of the dp Quattro series. This might share the same information that I talked last year, but please let me explain the idea.
The goal of the dp Quattro is to achieve the image quality with a tangible and realistic feel. We do not compete to increase the pixel number, but to acquire the higher image quality. My hope is to achieve the image quality which the medium sized film cameras used to have.
This is an off-topic episode, but Foveon's chief engineer likes to shoot with medium size film camera along with developing the image sensors. He is a unique guy. But unfortunately he is so busy lately that he has less chances to go out shooting. Anyway, to reach the goal, we SIGMA team and Foveon team collaborate each other with the same direction.
So how to acquire the “rich” image quality? We think it is necessary to get minute detail and stable detail for any kinds of objects. For DP camera, we think three factors are important. The first is the Foveon X3 image sensor. The second is the specially designed high quality lens. The third is the fixed, not interchangeable, lens which uses a lens shutter with less vibration and is aligned perfectly to the image sensor. An interchangeable lens camera relies its image quality on the accuracy of the lens mount as they align the image sensor to the camera mount and align each lens to the lens mount. On the other hand, this fixed lens camera is designed to adjust the matching of the lens and the image sensor from the center to the corners. So, the lens is physically “attached” to the image sensor to achieve the perfect resolution anywhere on the image. This is the characteristic of the lens fixed camera.
You might hear this many times, but please let me explain the Foveon X3 image sensor.
This is a normal Bayer sensor, which can not distinguish the color difference so it uses the RGB color filter to obtain color information at each pixel. Then, it compliment color information by calculation and make a full color image. This Bayer sensor is named after its inventor Mr. Bayer. We Japanese call him “Bay-yah,” but actually the pronunciation is “Buy-yah.” I heard his bereaved family also want to be called “Buy-yah,” anyway.
The next is the Foveon sensor. This sensor can capture different wave lengths depending on the depth of the silicon. Here is also what I talked last year. For normal color filter sensor, the green shares 50% of the pixels and the blue and the red share 25% each. On the contrary, the Foveon sensor can capture all RGB at any pixel. The reason why normal sensor spares 50% of the pixels for green is that human eyes are the most sensitive to green. The shorter wave length is blue, the middle is green, and the longer is red. Longer than red is infrared, and shorter than blue is ultra violet. Both human eyes can't see. As Foveon sensor can captures green at all pixels, if the number of the pixel is the same, Foveon has twice as high resolution as Bayer sensor. This is the principle of the Foveon sensor.
I'm surprised that a lot of audience here has Quattro cameras, as I have few chance to see someone with a Quattro camera on the road. Thank you so much.
The characteristics of the Quattro sensor is that on the top layer, it captures the detail of the object. It consists of 20 mega pixels. For colors, the top, the middle, and the bottom layers captures each color. The middle and the bottom layers are four times bigger than the top layer. As Foveon is three layer sensor, the data size is triple, and because of this, the processing, transmitting, and writing the data get slow. However, Quattro can solve this problem. Anther difficulty is that the wiring becomes complicated if we maintain the previous structure. The size of the photo diode becomes smaller and the amount of intake of light also becomes worse. By using the Quattro 1:1:4 structure, we do not sacrifice the intake, and could maintain the detail of Foveon X3.
These are the specifications of the Quattro sensor, which you might be aware of. The total number of the photo diodes is 29,400,000. It is comparable to the 39,000,000 pixels Bayer sensor in terms of the resolution.
With the Foveon sensor, whatever the object is, you could get the detailed image. I think the uniqueness of the photography is its reality, compared to other arts such as paintings. We would like to pursue the expression which has reality as if it exists in front of you. Therefore, it is necessary to avoid unnatural resolution, such as some area is detailed but other is not.
Here is a resolution chart taken by dp2 Quattro and 36 megapixel Bayer sensor. Both sensors' resolution is quite good, but on the Bayer sensor, the brightness resolution and the color resolution don't match. Therefore, there is false color. On the other hand, even it's structure is unique 1:1:4, Quattro's final image has no false color, that is, you can obtain a clear and clean image. This is the characteristic of Foveon, and with this image, we could achieve the “middle size quality.” These are the technological background of Foveon.
Before I started preparing this presentation, I asked some questions on twitter. Someone asked me “Is there any room to improve the quality of the Quattro sensor?” He has been using SIGMA camera for a long time, so he knows a lot. I felt it's a really critical question.
I have something to apologize for users. Sigma Photo Pro ver. 6.0 was released with dp2 Quattro, which is designed to develop Quattro file. Current version is 6.2. But previously, there are some errors and unoptimized processes on the SPP. For example, if you take a long exposure photo, the noise reduction works too strongly, so the detail had disappeared. I knew a lot of users wanted to take a detailed long exposure, but there was a bug on the SPP.
The second is that there was too much noise reduction on the shadow area. As I said, we want to make a high resolution camera with all kinds of objects. But there was a problem, though it's been fixed already. Finally, as we developed the new 1:1:4 structured sensor, which is totally different from its predecessors, we didn't understand everything and found there were ways to improve the image quality.
We have updated and keep updating the firmware to optimize the Quattro sensor. The next version SPP 6.2.1 is coming soon. We are planning to release this at the end of February. If you have a Quattro file, please download the new version, and re-develop it. I think a lot of users were worried about the Quattro image, but I'm confident that SPP 6.2.1 will satisfy you. We will keep working to improve the Quattro image and developing the appropriate algorithm.
Another question is “Can SPP processing become faster?” and “Are you going to sell a high speed PC for Quattro development?” Of course, we are not planning to sell a PC, but we understand that the slow processing speed is a problem. We have improved the speed quite a lot at the SPP version 6.2. However, it is also true that we have prioritized the processing quality rather than the speed. So we will focus more on the processing speed and put much effort to improve the usability of SPP.
I'm going to talk about the new product. The first is dp3 Quattro. As you know, it has 50mm lens, which is equivalent to 75mm on a full-frame. dp1 is equivalent to 28mm, and dp2 is equivalent to 45mm. In the series, dp3 is positioned as a middle telephoto lens camera.
Now, we have a new ultra-wide lens dp0. To say the truth, I'm a little worried about the naming of dp0. Someone wrote on twitter that if there will be a wider lens, will it be dp minus 1? The other said if there were 35mm dp, would it be dp 1.5? Even SIGMA couldn't release something 1.5, so I was wondering the name of dp0. But finally, as I'll talk about this later, the distortion of dp0 is almost zero, so it's wider than dp1, and zero distortion, dp0 represents these ideas.
Anyway, let's talk about dp3. As some of the DP3 users might notice, it has a really sharp lens. It can give a real DP quality image. The design of the lens is really classical. It consists of a double-Gauss master lens with a converter. Because its focal length is standard, we can choose a classic lens design, but we put the cutting-edge glass element into the classic design, and manufacture it at the most sophisticated factory. So it achieves the ideal performance of its design, and the yield rate is also high. A lot of users appreciate its image quality, saying it can take sharper image. Compared to all SIGMA lenses, this dp3 lens is one of the sharpest lens.
Updating to Quattro, we improved the AF speed. It's 20 to 30% faster than the previous DP3 Merrill. dp camera has a problem on focus at low light, but we also improved this issue. In addition, dp2 and dp1 Quattro's AF will improve with a new firmware, which will be released at the end of February. The exact day has not been decided yet though. So for Quattro users, please download the new firmware and try it.
The release of dp3 Quattro is in March. So, please think about it.
We also announced the 1.2 times conversion lens FT-1201 for dp3 Quattro. The equivalent focal length will become 90mm, and as it's a front converter, the F-stop is the same 2.8. The concept of this is the converter that doesn't deteriorate the image quality of the dp3 master lens.
This is the picture of the converter. As you can touch there, it's a big converter consisting of three groups and four elements. Here is the MTF chart. The left side is the dp3's master lens, and the right is the dp3 with the converter. The left side of the chart shows the center of the lens, and the right side is the corner. The red line represents low frequency, such as a thick big object, and the green line represents high frequency, small objects. From the center to the corners, both lines are flat and the performance is high.
With the converter, the performance at the center is the same. From the center to the corner, the performance is stable. But please check the far right. At the corner, the performance is actually higher than the master lens. I was wondering whether it is true, so asked to the designer. He said it's a coincidence. What a nice coincidence to get a better performance. I wouldn't say that all performance is better than the master lens, but the MTF chart is like this.
The picture of dp3 with the converter is here. The original lens consists of 10 elements, and the converter has four elements, which means 14 elements. As the body of the dp is thin, it becomes a front heavy camera. You might be worried about the balance, but please feel at ease, we have a LCD view finder. With this finder, the balance of the camera gets better.
Let me talk about the finder a little. The part which attaches to the camera body is called bracket. This part is made by aluminum machining, so it's really solid. The eyepiece lens is optical glass, which is as same as an interchangeable lens. The diopter adjustment uses an advance helicoid, so its feel is as sticky as a manual focus lens.
Here is dp3 with the converter and the view finder. As this becomes 90mm, it is better to attach the eyepiece to your left eye to avoid the camera shake. It will be a good portrait camera. But to say the truth, I can't understand what I'm doing on this camera, seeing its form. Anyway, please try using them.The converter is going to be released with dp3 next March, but I'm not sure whether it's the same day.
We announced the bottom grip too. We try matching the appearance to dp cameras. The top part is processed with alumite treatment which is the same process we used for dp's manual focus part. The knurling pattern is also as same as the dp's camera dial, so it matches really well. The screw is made of stainless steel, not aluminum, so it's durable.
Here is the sample. This is also made by aluminum machining, so you feel some heaviness and quality. If you attach it to the camera, it looks like this. Here is an image. Please try it. It will be available in April.
Then, let me talk about dp0. It has a really long lens, so when I check the customers' reaction on the net, they are saying “Isn't it 85mm?” Someone I met yesterday said “Are you putting a 10 times zoom lens into it, aren't you?” But to be honest, this is just an ultra wide angle lens. I'll explain the detail later.
The concept of dp0 is “zero distortion super wide camera.” Someone asked me on twitter saying “Why the lens is so long?” The reason is to achieve the dp quality. We try making the lens which could capture the minute detail of the object. The distortion is almost zero, and we corrected the chromatic aberration thoroughly.
We also want to put the light vertically into the sensor, so we achieved the higher telecentric performance. Because of these, the length of the lens became that long.
Here is lens composition. Actually, it is really fancy composition. We incorporated four FLD glasses which is resemble to fluorite and two SLDs, super low dispersion glass. The first large glass has aspherical surfaces on both sides. The final lens is also aspherical, so of all elements, normal lens is only three, and the rest is special glass, such as low dispersion or aspherical. So the composition is really gorgeous. Before I started preparing for this presentation, I talked with a designer of this lens, asking his intention and purpose. I said “Wow, it's really gorgeous and rich design,” and he said “Yeah, I've never expected that you would approve this composition.” I was surprised why he was saying at that time. Anyway, I'm happy that we could achieve high performance with this lens.
Here is a chart that explains the difference between normal glass and special glass. Through normal glass, the image blurs depending on the wave lengths. So there is an aberration. On the other hand, with special low dispersion glass, such as SLD or FLD, there is less dispersion, so less aberration.
By using these special glass a lot, the distortion of the final image is less than 0.5%. This is the least distortion of all dp cameras. Also, the chromatic aberration at the farthest corner of the image is one pixel. So the performance is really good.
The length of the lens contribute to achieve this high quality. As this is an ultra wide lens, you have to intake the light from a wider angle, and the light has to reach a large image sensor. If you bend the light drastically, it is difficult to avoid the appearance of the aberration. Therefore, we try bending the light gradually and carefully. Like an “In box daughter” (We Japanese call an only daughter brought up carefully like this), we avoid the light becoming delinquent, and carefully bend the light element by element. If you bend the light suddenly, there will be an aberration, so we have to correct it too, by sending her to a correctional institution. That's why the length gets longer like a telephoto lens.
Someone might wonder why we are working so hard to increase the lens performance. A lens distortion and a chromatic aberration can be fixed by the camera because it has a prime lens. However, this is also the same idea of the dp quality: to get clean and fine details. It's easy to understand the correction of the distortion. As the corners are smaller than the center, you have to enlarge these corners, then it will lose the detail. Although the software correction is more powerful and excellent, the restoration of the detail is impossible. We have to rely on the lens and the image sensor to get the detail. If you fix the distortion, you will lose the detail.
If it is a Bayer sensor, the correction is not distinguishable. On the contrary, as Foveon can get one pixel image, you could notice corrected area easily.
If I say so, you would think that it's not so difficult to correct the chromatic aberration as it is only the diffusion of color. Of course you could fix the aberration depending on the wave lengths, but it is not true.
Here is a simulation of dp3 and our classic 14mm F2.8 lens. The performance is terrible, but it's already discontinued, so it might be OK to show here. You can see the diffusion of color.
This is an excerpt of green light; the middle wave length. As you can see, even it's the middle wave length, the light is diffused. You can only correct the color with the chromatic aberration correction, and still you can't get the detail. Even the dp0 has an aberration, but no worries. It's an enlarged image, so the aberration is really small. Anyway, if you say you don't need to be worried about the chromatic aberration because you could fix it later, it's not true. With a less aberrated lens, you could acquire a sharp, fine detailed image. The goal of dp0 is this, so it has a long, large lens.
Here is dp0. It has such an extremely large lens that no one can believe it is a wide angle lens camera.
We also made a view finder of 21mm equivalent. I have no idea how many people would buy this, but anyway, we made it because it's necessary. It has a high magnification lens, and the eye point is relatively long, so you could see well with a naked eye or with glasses. The body is aluminum machining, and glasses are optical lens, so the quality is excellent. If this sells well, we could pursue the next project easily, so please think about it.
I haven't decided the release date, but I would like to start selling dp0 at the beginning of summer. With dp0, the dp Quattro series consist of cameras with ultra-wide angle to middle telephoto lens. This is dp0: 21mm equivalent, dp1: 28mm, dp2: 45mm, dp3: 75mm, dp3 with a converter: 90mm. Actually, it's very different, so you could call this dp4. As a whole system, they cover nice focal lengths. If you bring everything at once, it's a little bulky, anyway. But I believe we could make a good system.
Let me announce here about firmware update for dp1 and dp2 Quattro. It is scheduled at the end of February. As I mentioned, the AF at low light will improve. We add new aspect ratio, 7 : 6, new image size “super low,” a very small size. You could minimize the time lapse timing to 4 seconds. You could add and check the copyright information. You could use translucent frame when you change the aspect ratio. With this function, you could see an object outside of the frame. This translucent frame might remind some of you our SD9, which was also a “crazy” camera.
Someone asked me on twitter whether there will be a successor of SD1. Sorry, there is no surprise today, so I couldn't show anything. But we are working on the SD. Many people ask me about mirror less dp, but for us, the successor of the SD is a top priority. Someone named it freely like SD1 Quattro, but the name, the spec, nothing has been decided yet. We are working hard, so please give us a time.
We are often asked about Full Frame Foveon, but basically it's no comment. If you ask me if there will be a FF Foveon next month, I'd say no. But it doesn't mean there will be no FF Foveon in the future. I can't say yes or no.
What I want to say is that there are a lot of challenges when we make FF Foveon. Even with Quattro 1:1:4 structure, the data size will become huge. The data processing time and writing time will be a problem. More difficult problem is lens. The bigger the image circle is, the more difficult to design the optics.
At present, it is true that the APS-C size Foveon is well-balanced. As you saw, the normal Bayer sensor can get only half resolution as Foveon can. So for Bayer sensor, there is only one solution to increase the resolution: increasing the sensor size. Because there is a threshold to minimize the pixel pitch.
I believe that with a smaller sensor, Foveon is the only way to achieve the higher resolution, and APS-C Foveon can be comparable to FF Bayer sensor in its performance. If we make FF Foveon, considering the data size and other things, it will behave something like middle sized digital back. Please keep in mind these things. But basically for FF Foveon, it's no comment.
Finally, let me talk about the Foveon X3 technology. We are aware of its weakness, such as high ISO performance. As only we and Foveon are developing the sensor, it is difficult to increase the speed. But still, under some circumstance, Foveon can take a unique, special image. I believe it's a Foveon's raison d'etre. To respond to the customers' demand, we'd like to keep improving the sensor.
Another reason why we are working on Foveon is that I believe it has some historical significance. Some of you might know that the first color photography was French Lumiere brothers' autochrome. It was invented in 1903. They are also famous as the founder of the movie. These are the autochrome pictures taken at the beginning of the 20 century. The Lumiere brothers were raised in Lyon, France. I had a chance to go to Lyon last year, and spared some time to visit the Lumiere museum which exists in the suburbs. There were many pictures taken at the beginning of the 20 century, so I was really impressed. There were also some pictures taken in Japan during that time. It was really interesting.
So how they could make an autochrome image. They dyed potato starch into red, green, and blue, and put it onto the glass board. Then they exposed the glass. You might guess what I'm going to talk next, but anyway. Later, Kodak invented a color film, which captures colors vertically. So the color film era had started since then. The normal image sensor captures colors horizontally, though its performance has been improved significantly. However, the basic concept of the Bayer sensor is as same as the Lumiere brothers' autochrome. On the contrary, Foveon takes the “vertical” route which Kodak had broken through. We believe that this vertical approach has technological future, so we pursue this route. We have lots of challenges, but are working hard with hope.
We have a lot of ideas to implement into the Foveon sensor. It takes time to make everything feasible, but we SIGMA and Foveon are working really hard to improve the Foveon sensor. Please expect our future products and support us.
Thank you for listening.
Quentin_Bargate
Well-known member
Thank you, Hulyss, for that very helpful translation.
Kazuto Yamaki says:
"We have updated and keep updating the firmware to optimize the Quattro sensor. The next version SPP 6.2.1 is coming soon. We are planning to release this at the end of February. If you have a Quattro file, please download the new version, and re-develop it. I think a lot of users were worried about the Quattro image, but I'm confident that SPP 6.2.1 will satisfy you...
...we SIGMA and Foveon are working really hard to improve the Foveon sensor"
My guess is they have been working hard on the issues early adopters of the Quattro cameras have identified. Whether they have made enough progress to satisfy critical users - such as premature highlight clipping, or mottling in shadow areas - is unclear. But let's give them a chance.
Kazuto Yamaki says:
"We have updated and keep updating the firmware to optimize the Quattro sensor. The next version SPP 6.2.1 is coming soon. We are planning to release this at the end of February. If you have a Quattro file, please download the new version, and re-develop it. I think a lot of users were worried about the Quattro image, but I'm confident that SPP 6.2.1 will satisfy you...
...we SIGMA and Foveon are working really hard to improve the Foveon sensor"
My guess is they have been working hard on the issues early adopters of the Quattro cameras have identified. Whether they have made enough progress to satisfy critical users - such as premature highlight clipping, or mottling in shadow areas - is unclear. But let's give them a chance.
darr
Well-known member
I hope to tryout the DPQ0 as the 21mm lens is the closest to a favorite format for me (20mm). Hoping they get the software issues resolved. I was using the latest SPP this past week for Merrill files and noticed considerable speed increase on my PC, so some improvement has been made.
Quentin_Bargate
Well-known member
A street artist known as Paul Don Smith regularly adds new work to the streets of East London, often overlaying earlier work and other stuff. Here are some recent examples taken on the DP2Q. These were developed using the latest SPP 6.2.1. I have also now upgraded to the latest firmware.
Another bug of sigma dp2 Quattro:
Have a look at blue branches on a cloud on the right side
and zoomed on a cloud:
DP2Q2370-BLUE BRANCHES by Dimy4, on Flickr
Have a look at blue branches on a cloud on the right side
and zoomed on a cloud:
DP2Q2370-BLUE BRANCHES by Dimy4, on Flickr
Quentin_Bargate
Well-known member
I posted about this issue on the DP1 Quattro thread,
http://www.getdpi.com/forum/other-cameras/52488-sigma-dp1-quattro-shots.html
At #2, and I also reported it to Sigma. Your example shows it even more clearly. It shows itself most obviously with fine tree branches shot against an overxposed sky. The effect is absent from the Merrills.
Given the latest firmware and software make no difference, I assume it is hardware related. In any event, it is a yet another issue to contend with if shooting landscapes or high contrast subjects with a Quattro.
Stoneage
Member
Is it possible to make the RAW available? That would be very interesting.
Stoneage
Member
Thanks
As expected, there is actually more information in the X3F, but Sigma (SPP) has problems to get the most out of the files.
As somebody else mentioned, they seem to get all luminance information only from the top layer and if it is blown, then the data is lost, even if the other layers still have informations.
With RAWDigger, the critical part of the image still seems to have useful data:
Quentin_Bargate
Well-known member
Impressive. Perhaps there is a future for these cameras if Sigma can utilise luminance info from the lower layers in extremis. I assume they do not do so at the moment because the lower levels are much lower resolution (unlike the Merrills).
Wow! i didn't know about this tool. Thanks for sharing this
Yes, probably... However it looks really strange to me now, after having seen how better the image could be using info from other layers
octagone
New member
On the occasion of a pro salon which took place in Paris for two days, I had the opportunity to try the dp3q and especially dp2q.
In case its interest to some of you, here are some pictures, processed from RAW and TIFF.
Dp3q - 800 iso.
Full res:
https://flic.kr/p/uiA4B9
Full res 39Mpix:
https://flic.kr/p/uY2rLA
Dp2q - 1600 iso:
Full res:
https://flic.kr/p/vfVuXD
Dp2q - 200 iso:
Full res:
https://flic.kr/p/vdhsrs
Full res 39Mpix:
https://flic.kr/p/vdhAsN
Dp2q - 400 iso:
Full res:
https://flic.kr/p/uRF15d
Full res 39Mpix:
https://flic.kr/p/vfzLVp
In case its interest to some of you, here are some pictures, processed from RAW and TIFF.
Dp3q - 800 iso.
Full res:
https://flic.kr/p/uiA4B9
Full res 39Mpix:
https://flic.kr/p/uY2rLA
Dp2q - 1600 iso:
Full res:
https://flic.kr/p/vfVuXD
Dp2q - 200 iso:
Full res:
https://flic.kr/p/vdhsrs
Full res 39Mpix:
https://flic.kr/p/vdhAsN
Dp2q - 400 iso:
Full res:
https://flic.kr/p/uRF15d
Full res 39Mpix:
https://flic.kr/p/vfzLVp
Last edited:
octagone
New member
Dp2 quattro:
New version 1600 iso.
https://flic.kr/p/vV1XQ8
New interpolation tests:
39Mpix:
https://flic.kr/p/vTn7gT
50Mpix:
https://flic.kr/p/vAKkhU
A photo with a forum dpreview member sent me a raw:
http://www.dpreview.com/galleries/165712628/photos/3247225
With a bit more sharpness and detail in the shadows:
http://www.dpreview.com/galleries/165712628/photos/3247574/falun-013-rt-bis
New version 1600 iso.
https://flic.kr/p/vV1XQ8
New interpolation tests:
39Mpix:
https://flic.kr/p/vTn7gT
50Mpix:
https://flic.kr/p/vAKkhU
A photo with a forum dpreview member sent me a raw:
http://www.dpreview.com/galleries/165712628/photos/3247225
With a bit more sharpness and detail in the shadows:
http://www.dpreview.com/galleries/165712628/photos/3247574/falun-013-rt-bis