Canon 5D Mark II

[Ben Rubinstein]

I saw the same in the Nikon files I've seen Amin but I've never had a RAW to process, I've asked a member here to upload a D700/3 file shot with the 24-70 for me to have a look for myself, it's no use comparing others observations or looking at their processing, to see what a file looks like - for you - you need to process the file in your usual RAW converter with the look that you like, then compare.

 

[mazor]

Actually, I had hoped Canon would have heard this "aggressive" comment from consumers ... and used this camera as a course correction.

I cannot tell you how disappointed I was in the file quality of my 1DMKIII and IDsMKIII ... both the cameras were great in every respect except the thing you actually use them for ... images. A friend that bought my MKIIs made me jealous, and I longed for those cameras back.

I actually believe that post quoting an unnamed Canon source, whether it's legitimate or not. I think they could blow the doors off the competition because they have the resources ... but succumbed to the marketing ploy of relentless mega pixel race combined with the high ISO race forced by Nikon ... which appears to be a lethal cocktail so far as IQ is concerned.

It will be VERY interesting to see how Nikon handles any high resolution DX flagship camera. I wonder if consumers would accept a sub 20 meg sensor with larger photosites and a rational balance of high ISO and IQ? Or will marketing drive that also?

I think there is a possibility that nikon may use the new full frame 24MP sensor from the A900 in the new D3x. using the Expeed processor instead of the bionz should give a different result.


Mazor

 

[asabet]

I saw the same in the Nikon files I've seen Amin but I've never had a RAW to process, I've asked a member here to upload a D700/3 file shot with the 24-70 for me to have a look for myself, it's no use comparing others observations or looking at their processing, to see what a file looks like - for you - you need to process the file in your usual RAW converter with the look that you like, then compare.

Ben,
I'd be happy to upload some D700 RAWs for you. If you want them specifically taken with the 24-70, then I have to wait for my 24-70 to arrive . Otherwise I can shoot something with another good lens and upload the RAW for you. Anything in particular you'd like to see subject-wise?

 

[mazor]

There are a few references here to 'plastic' images. I don't think this means some sort of three-dimensional effect.

Could someone please explain to me just what a 'plastic' image is? And while you are at it, just what is 'micro-contrast' - another term I keep seeing and not knowing what it is.

Apologies for being so dense. Thanks.

hi,

For me a plastic image is something like a water colored painting. it is definitely not a 3D effect! basically it is noise processing built into the camera. If the processing is too much, colours smudge together transforming a textured surface into a smooth unrealistic surface. Usually we see this effect more at high isos but in this case we can see it in the 5D mk ii at ISOs as low as 100.

Mazor

 

[Ben Rubinstein]

Ben,
I'd be happy to upload some D700 RAWs for you. If you want them specifically taken with the 24-70, then I have to wait for my 24-70 to arrive . Otherwise I can shoot something with another good lens and upload the RAW for you. Anything in particular you'd like to see subject-wise?

PM Sent!

 

[fotografz]

There are a few references here to 'plastic' images. I don't think this means some sort of three-dimensional effect.

Could someone please explain to me just what a 'plastic' image is? And while you are at it, just what is 'micro-contrast' - another term I keep seeing and not knowing what it is.

Apologies for being so dense. Thanks.

I know you can get that "blow-up doll" skin with any camera by over processing in post. The problem arises when that effect is introduced by aspects of the camera before it even gets to post work. :banghead:

IMHO, this is partly the consumers fault ... the demand for those smooth looking files where 35mm DSLRs are suppose to rival the tonal gradations of a Medium Format file ... plus the drive to cram more and more mega pixels into the same small space ... coupled with attempts to jack up ISOs while suppressing noise and moiré ... have led to compromises in IQ :thumbdown: ... whether they are worth it is a very personal evaluation.

For example, my 10 meg M8 files or DMR files of people are more "human looking" than those I got from my 22 meg Canon 1DsMKIII.

The file links posted in this thread of the young girl exemplify that IMO. But they are jpegs, so it's difficult to know how much of the effect was introduced in processing ... and if it was introduced, why?

Those pics are worth downloading and looking at @ 100%. Look at the skin. It's waxey. The lips are a like drug store wax lips I used to buy as a kid. It's exactly like what I was getting from the 1DsMKIII ... a look I personally do not like ... but others may not find objectable, and some may even prefer.

To each his or her own.

 

[mazor]

fotografz. I am in total agreement. with you here. DMR shots not only have the 3D effect on focused subjects but also, the image looks really natural. I suspect the DMR CCD sensor delivers a much cleaner default image, whereas CMOS does not, and in order to get an acceptable image (low noise) they apply fancy algorithms in the "digic" to combat noise before even it hits the RAW file.

My older canon 10D with digic i seems to reveal much more natural images in my opinion compared to my 5D with Digic ii. It seems with every generation of digic, Canon is going trying to combat more and more noise as the megapixel increases and the result is an increased intensity of "plastic" or "water color" effects very visible at low ISOs even.

It is interesting to note that images I have seen from the d700 and the d3 are much more natural, and seems to have a more filmic and more natural appearance, similar to that of the DMR which in my opinion tops all in this area.

Mazor

 

[LJL]

Well, after looking at those two images once again, I am still not quite sure what folks are describing as "plastic" in the look. The lips, as Marc points out, do have a waxy look, but I think that is the lip gloss that the model is using. For the skin, it is really hard to tell what is going on, as this fairly young woman (girl) appears to have a fine powder or something smoothed over her already tiny pores, thus rendering things somewhat smooth. If you look at the eyes, lashes and eyebrows, there appears to be a fair amount of detail that is not getting as smoothed out as what is visible on the skin and lips. Same with the hair at the hairline.

I am not trying to defend the shots from this camera, and since they are JPEGs, it is entirely possible that whatever in-camera processing was applied does blur some of the finer detail. It just seems that the application is not uniform over the entire focused area, hence my questioning what folks are looking at when they say things look plastic. I would think that all areas would exhibit the same gentle smearing if it was the processing. The iris in the eyes does not seem to be so smeared/smoothed.

Basically, I agree with some comments folks have made about sacrificing gritty image detail for a smoother look. My M8 files with almost any lens carve out more pores and fine details, unless they are covered with even a bit of make-up. Just offering that until we start seeing some RAW files that we can process ourselves, puffing up or tearing down the image quality capabilities of this, or any camera, is a bit premature. If folks want to use these shots as definitive results, so be it, but I think a lot may be missed.

The CMOS sensor has never quite been able to capture the same sort of resolved detail as some CCD sensors, but we are not always sure of just how much the AA filter, reading the Bayer array, or other in-camera processing caused those differences. So, if somebody is looking for a 21MP FF DSLR, does shoot RAW, and has a good processing routine, I think they may not be too disappointed with the amount of detail that this camera captures. Personally, I think it is still too early to tell just how good or bad this new camera is.

My comments about DOF had NOTHING to do with "3D look" as may have been brought into this discussion. I was commenting that the first file linked by LCT had an extremely shallow DOF to the point that the eyebrows were OOF just as the eyelashes were starting to come into focus. And the model' right eye is not fully inside the plane of focus, from what I can tell. Therefore, most of the skin one is looking at is OOF, so it should have a smoother look that some see as "plastic". Just my thoughts on this. In the second shot, more things are in focus, and when you look at things at 100%, you can see some of the details, especially around the base of her nose. So, folks may still think all of it has a "plastic" look to it, and that is fine. I just wanted to understand what they were looking at to make those conclusions. I agree that the razor edge sharpness seems a bit smooth, but not sure what is glass, what is sensor, what is processing, what is lighting, what is shooter influenced. Again, not defending the shots from this camera, just trying to create a bit separation and clarity in parts of the discussion so that folks can draw their own conclusions, and not walk away with an impression that blankets things too positively or negatively. I still say we need to wait for some RAW files, more shooting info, and then process them the way we would prefer. I know I have to process my 1DsMkII files differently from my 1DMkII files, and those are both different from my M8 files, but I have found that all can deliver very good results, just not so easily in a "one process fits all" manor.

LJ

 

[Ben Rubinstein]

The man has a point if we are only talking about the jpg's. There are some portrait samples from the Sony online, I had a look, first thing I said when viewing was 'that camera is butchering the jpg's!' Second comment was that the Zeiss 24-70 is very nice (very smooth focus to OOF rendition)! The fine detail in the faces was horribly over smoothed.

I think what Marc is saying is that the overall 'look' from the more recent canon files is plasticky, many said that this was from the 1Ds mkII onwards as commercial photographers often complained about the plasticky rendition of facial tones on the mkII relative to the original 1Ds, it was dogma on the old RG forums. I personally love my 5D's files but time and time again I see Nikon facial colour that creams my 5D. I just can't get that film like colour in the faces and I've been processing 5D RAW files for 3 years, the 1Ds, 10D and D60 before that and ran a photo lab before that! I have seen some custom C1 profiles that look really good but I use an ACR workflow.

I had expected the release of the Adobe colour profile editor for RAW to provide a flow of custom profiles from the names that make the C1 profiles, stuff that we could really get excited about. It seems to have been rather quiet on that front though which is a shame, I'd been hoping for my 5D files to take on a new meaning for facial tones.

 

[robmac]

There was something about those 1Ds Mk1 files...

 

[LJL]

Ben,
Really did not want to mix apples and oranges here. The colors you are talking about in Canon v. Nikon are completely different than the "plastic" smoothing issue. I still think Canon does not get skin tones perfect, and I personally have not liked them myself. I think Canon also screws up the reds and yellows to some degree, but some of that can be fixed in process. Nikon files tend to have a more "cool" or slightly blue cast to them, much like what I used to like shooting Ektachrome film long ago....always about a third stop underexposed and slightly cool.

Those color differences are separate from the sharpness/smoothness/smearing issue that is usually attributed to things looking more like plastic. It is interesting to note that almost all portrait retouching, with the exception of gritty B/W work, looks to smooth the skin, close the pores, remove the imperfections. However, there is a big difference between having that done in the camera, versus being applied in post by the photog.

As an example of how things can look radically different, take a look at some of the shots folks have posted from the Lighting Workshop (in both the MF forum and in its own Workshop Forum thread). So many things going on there with respect to skintones. Some models have make-up that looks like it was applied with a putty knife, while others have less. The lighting appears more side-angled which picks up more detail than beauty dish type lighting the softens and rounds things. The skintones are all over the place, even though folks are using the exact same lighting set-ups, so individual WB is different for different tastes. Some folks are using MF backs with no AA CCD sensors, others appear to be using Nikons and other CMOS sensors with varying AA filter strengths. If anyone looked at just one shot to make statements about how something was performing, I am not sure I would feel comfortable with those conclusions, since given the same models and the same lighting, we are seeing vastly different results, some attributable to the sensor, some to the glass, some to the processing or lack thereof in some cases.

Just thinking about how we analyze things and the conclusions we sometimes draw from those analyses.

LJ

 

[fotografz]

Well, after looking at those two images once again, I am still not quite sure what folks are describing as "plastic" in the look. The lips, as Marc points out, do have a waxy look, but I think that is the lip gloss that the model is using. For the skin, it is really hard to tell what is going on, as this fairly young woman (girl) appears to have a fine powder or something smoothed over her already tiny pores, thus rendering things somewhat smooth. If you look at the eyes, lashes and eyebrows, there appears to be a fair amount of detail that is not getting as smoothed out as what is visible on the skin and lips. Same with the hair at the hairline.

I am not trying to defend the shots from this camera, and since they are JPEGs, it is entirely possible that whatever in-camera processing was applied does blur some of the finer detail. It just seems that the application is not uniform over the entire focused area, hence my questioning what folks are looking at when they say things look plastic. I would think that all areas would exhibit the same gentle smearing if it was the processing. The iris in the eyes does not seem to be so smeared/smoothed.

Basically, I agree with some comments folks have made about sacrificing gritty image detail for a smoother look. My M8 files with almost any lens carve out more pores and fine details, unless they are covered with even a bit of make-up. Just offering that until we start seeing some RAW files that we can process ourselves, puffing up or tearing down the image quality capabilities of this, or any camera, is a bit premature. If folks want to use these shots as definitive results, so be it, but I think a lot may be missed.

The CMOS sensor has never quite been able to capture the same sort of resolved detail as some CCD sensors, but we are not always sure of just how much the AA filter, reading the Bayer array, or other in-camera processing caused those differences. So, if somebody is looking for a 21MP FF DSLR, does shoot RAW, and has a good processing routine, I think they may not be too disappointed with the amount of detail that this camera captures. Personally, I think it is still too early to tell just how good or bad this new camera is.

My comments about DOF had NOTHING to do with "3D look" as may have been brought into this discussion. I was commenting that the first file linked by LCT had an extremely shallow DOF to the point that the eyebrows were OOF just as the eyelashes were starting to come into focus. And the model' right eye is not fully inside the plane of focus, from what I can tell. Therefore, most of the skin one is looking at is OOF, so it should have a smoother look that some see as "plastic". Just my thoughts on this. In the second shot, more things are in focus, and when you look at things at 100%, you can see some of the details, especially around the base of her nose. So, folks may still think all of it has a "plastic" look to it, and that is fine. I just wanted to understand what they were looking at to make those conclusions. I agree that the razor edge sharpness seems a bit smooth, but not sure what is glass, what is sensor, what is processing, what is lighting, what is shooter influenced. Again, not defending the shots from this camera, just trying to create a bit separation and clarity in parts of the discussion so that folks can draw their own conclusions, and not walk away with an impression that blankets things too positively or negatively. I still say we need to wait for some RAW files, more shooting info, and then process them the way we would prefer. I know I have to process my 1DsMkII files differently from my 1DMkII files, and those are both different from my M8 files, but I have found that all can deliver very good results, just not so easily in a "one process fits all" manor.

LJ

Clarity: it looks waxy and inhuman like a cadaver. It appears to be an artificial smoothing for some reason. Look at some fashion shots with lip gloss ... doesn't look like that, trust me.

Nothing in the entire shot is sharp ...except when viewed small. I'd reserve any premature judgement IF (big IF) my previous 3 Canons didn't deliver the same look. So I'm skeptical based on experience and will remain so until fresh info and proof it ain't so is forthcoming.

 

[ptomsu]

Interesting how a new product is evaluated without having real data to do so ....

I find all these discussions about if, when, perhaps, what, etc useless.

Why not wait till there are real RAW data available for comparison and then try to judge the IQ of this beast?

 

[Robert Campbell]

My understanding of "plastic" is an image quality that looks overly processed. Noise reduction and AA filter-induced blur rob natural texture, contributing to a plastic look. Well done sharpening can help emphasize the natural texture, but only so much.

Microcontrast is a property of the lens, referring to the ability of the lens to discriminate between adjacent small areas of similar tonal value.

hi,

For me a plastic image is something like a water colored painting. it is definitely not a 3D effect! basically it is noise processing built into the camera. If the processing is too much, colours smudge together transforming a textured surface into a smooth unrealistic surface. Usually we see this effect more at high isos but in this case we can see it in the 5D mk ii at ISOs as low as 100.

Mazor

Thanks chaps. I'm still not entirely clear about this - perhaps because of 'plastic' having several rather different meanings.

Would I be right in thinking that 'plastic' is what other people call smearing?

Could anyone point me to where I can compare 'plastic' and 'non-plastic' images? And somewhere that shows me just exactly what this 'micro-contrast' is, and how it differs from 'contrast' and acutance?

Thanks

 

[mazor]

There was something about those 1Ds Mk1 files...

yes there is something about the 1Ds MK i files. I believe it is the Digic i processor that is responsible for this. Mabe someone could find a way to put this processor into the 5D or 5D mk ii, but then I would expect the "superior" low noise images we are so used to to diminish quite quickly.

as i said earlier, with every generation of digic comes improvements in speed, etc but also a so called "improved" noise reduction which each time has to be enhanced to combat the increased noise that higher megapixel sensors produce.

In my opinion, Canon should be developing a totally new design sensor rather than milking an existing design, by just cramming more pixels into a smaller space and trying to make up for it by using gapless microlens, etc to try to help reduce noise. I was really expecting this for the 5D mk ii.

Mazor

 

[mazor]

Robert Campbell,

I am not sure exactly where you can do to compare this phenomena. But I suspect if you google "DMR vs Canon" or similar you will find some interesting results.

Alot of people blame the AA filter found in both Canon and Nikon cameras, which is absent in both the Leica M8 and DMR, but that does not explain how Nikons latest developed D3 and D700 sensor manage to maintain a much more natural look which is less watercolored when pixel peeping than Canon.

MAzor

 

[mazor]

Interesting how a new product is evaluated without having real data to do so ....

I find all these discussions about if, when, perhaps, what, etc useless.

Why not wait till there are real RAW data available for comparison and then try to judge the IQ of this beast?

As an anxious 5D owner, one would be naturally curious about the 5D mk ii as a potential upgrade, and also people are discussing this topic on whether to get the new 5D mk ii or get the older 5D for a much cheaper price.

You are correct, we have no real RAW data to play with, and also it i possible that the firmware used may have been beta. We are just judging on JPGs on what we get to see, and hope that mabe Canon may be listening to these comments on how to remedy the "plastic" look if at all possible before official release.

Mazor

 

[johnastovall]

I'll find out how good it is. I've got one on order. I don't expect to be disappointed but then to me a camera is just a way to paint and draw with quanta, not capture reality.

Ah, ISO 25600 and an f/1.0 lens hand held by moonlight.

 

[mazor]

I'll find out how good it is. I've got one on order. I don't expect to be disappointed but then to me a camera is just a way to paint and draw with quanta, not capture reality.

Ah, ISO 25600 and an f/1.0 lens hand held by moonlight.

Nice work, lookforward to seein some good samples. ISO25600 even on d3 does not look overly hot though o don't expect miracles here.

Mazor

 

[Jonathon Delacour]

And you know the thing that they did change that gives me interest? The creative possibilities of using that 1080p video stream, and intermixing high quality stills into the flow. Go price a full frame CMOS sensor HD video camera, and get an eye full if you think this camera is expensive. I have. Many thousands of dollars more. Few of us have already seen a glimpse of the future, and my brothers, it is all about the convergence of still and video. Seen any of the new LED billboards yet? You will. I'd guess within two years, the old paper & plaster "signs" will all be gone here in the USA. Replaced with moving images, both rotating stills and actual video capture. What this means for professional photographers astute enough to grasp the implications, a huge windfall of new work with very few presently able to do it. The signs are all there, and I for one, am all over it. Save for one thing.... the cost of the camera. The only other full 35mm frame HD 1080p capture device I have found is the RED. And it is cheap at only $18,000 out the door. Without a stinking lens! Those can run more than the body, much more in fact. At $2,700, this camera is an absolute steal, provided of course that the actual image quality is up to the claims on the spec sheet. Let us also not forget what kind of magic mounting the now relatively inexpensive Leica R glass collection onto the front of a 1080p video capture device could reveal.

Given this is primarily a forum about still photography, I guess I shouldn't be surprised that Chuck Jones' comment about the convergence of still and video has (so far) been totally ignored. Yet, as much as I love still photography, I can't help thinking that Chuck is correct. Not that video will replace still photography but that the merging of the two disciplines will produce a third.

Two essays at Luminous Landscape are relevant to this issue. The Convergence of Still Photography & Video fleshes out Chuck's remarks while Understanding Video: A Video Primer for Photographers provides a decent introduction to both current video technology and the differences between still and motion recording.

While the RED ONE does indeed cost US$17,500, RED founder Jim Jannard has promised to announce the details of the RED DSMC (Digital Still & Motion Camera) at the end of this year -- for delivery at the end of 2009. In this thread on the REDUSER forum, Jannard says that the DSMC is "strategically targeted at the DSLR space. As Nikon and Canon release their 720P and 1080P, respectively, DSLRs with video capture... RED has a more advanced view of the future. We look forward to rapidly pushing the "big guys" along in feature sets and capabilities."

The 720P and 1080P cameras he mentions are the Nikon D90 and the Canon 5D Mark II respectively.

In another REDUSER forum thread, Jannard writes:

Given the announcements and release of still cameras from "others" that now shoot video, RED is excited to enter this game. From our vantage point, it is a lot easier to enter the still market from a motion background than visa-versa. The biggest issue that needs to be solved by the still capture group is skew... slow read-reset of CMOS imagers. This "typical" CMOS issue shows itself by moving the camera during motion capture. It is seen as "jelly movement". Red has overcome this issue with a rapid read-reset CMOS sensor program. The Monstro Mysterium sensor is the fastest read-reset CMOS known to man enjoying the same motion characteristics as a film camera.​

He also points to this D90 video sample which shows the "jelly movement". It's atrocious.

This page on the Canon Japan website has eight sample movies. Interestingly, in all but one the camera is locked down on a tripod while the other is a 360-degree pan using a fisheye lens (which one assumes would minimize wobble). In this 5D Mark II video on the Canon Korea website, the camera is also locked down. DPReview has two large video files one one of their Canon 5D Mark II preview pages. About these samples, a REDUSER forum member writes:

You guys need to watch the video samples on DPREVIEW.COM! I cannot believe Canon is using these as marketing materials.

The video looks very over-sharpened with horrible colorimetry - almost like HD cell phone video!! and I can see the same wobble wobble that the Nikon has.​

As Jannard points out, "It seems very interesting to me that every shot is a lock-down shot. The camera is not moving in any sample footage."

In 1964 Andy Warhol and Jonas Mekas made Empire, a silent black and white film consisting of 6 hours and 40 minutes of continuous real time 16mm footage of the Empire State Building. When projected at 16fps (as Warhol specified), the film runs for about 8 hours and 5 minutes.

The Nikon D90 seems best suited for this style of movie making. It'll be interesting to see some video footage shot by actual Canon 5D Mark II users, rather than the uninformative samples that Canon has provided.