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!

Sony 850 NOISE TEST

Guy Mancuso

Administrator, Instructor
Thought I would process these 1600 files . They are underexposed so give them a little credit since they are at least a half stop darker in exposure than they should be



 

mjm6

Member
Guy,

I think you should try L3 on this camera for high speed conversions. The improvement from L2 to L3 was substantial, and it has shown significant benefit on the M9 and also on the a900 that I have.

Funny thing though, when I'm shooting for B&W at high speed, I like the grain that the a900 produces, so sometimes I'll use the older rendering engine to keep the noise in the image. For color, I can't see myself not using the L3 rendering engine.

If high speed color is the primary purpose of the camera, download L3 and do a few comparisons with the C1 images you have. Or, if you want, post them somewhere and I'll download them and crunch them through L3 for you. The improvement is not subtle.

---Michael
 

Guy Mancuso

Administrator, Instructor
Thanks Michael look above in one of the posts i have a link to all the raws. LR is not on my system and it may never get there either. LOL

I'm a C1 fanboy total admit it and I also shoot a Phase back so I'm in for the long haul. Please do some raw conversions would love to see them though. My theory is this be a expert at one of these programs than be so so at both.
 

Guy Mancuso

Administrator, Instructor
Also we need to be careful as well sure i could knock the noise down more but it could be at a cost to detail. LR by default shows better noise reduction but watch the detail as it may slip.
 

mjm6

Member
Guy,

I know you are a C1 person, which is why I figured you may have not tried L3. I agree that ultimately, it is better to learn a product and then use it to best effect rather than tilting at the windmills of 'the best' at whatever particular aspect of raw conversion you wish to use.

In my case, I've been using LR since before it was LR, (back in the days of Raw Shooter Pro), so I am comfortable with it.

I'll try to download a few and run them through LR3 tonight and then post a few image snips if that's OK with you.

---Michael
 

mjm6

Member
Guy,

I noticed you mentioned that the MFDB acts 'more like a 15MP sensor', and I think this is an interesting perspective conceptually.

I know what you are saying, but I think of it in the exact opposite manner... A sensor with an AA filter installed will have a reduced resolution capability that ultimately makes it behave like a lower resoluton sensor in some ways (the lpmm capabilitiy of the lens/sensor combination will be reduced, but the overall magnification capability of the file will not).

Rather than stating that the MFDB is capable of performing to an equvalent 35mm sensor cability, I think of it as how a 35mm sensor acts more like a lower MP MFBD sensor, for example. I do this because I believe that a non-AA filtered sensor is the truest reflection of the performance of a propduct at that megapixel level.

The M9 is a good example of this, as it out performs the a900/a850 in some ways considerably, and is 'only' 18MP versus the 24.5MP of the Sonys. In unfiltered MP performance, I feel the Sonys run around 13-15MP compared to the M9, although some of the actual performance may also be related to the lenses (in my case Leica R on the Sony and Leica M on the M9).

I sure do like the way the Sony cameras perform, though. After 25+ years with Canon, I switched over last summer, with no regrets so far.


---Michael
 

jonoslack

Active member
Michael
The M9 is a good example of this, as it out performs the a900/a850 in some ways considerably, and is 'only' 18MP versus the 24.5MP of the Sonys. In unfiltered MP performance, I feel the Sonys run around 13-15MP compared to the M9, although some of the actual performance may also be related to the lenses (in my case Leica R on the Sony and Leica M on the M9).

I sure do like the way the Sony cameras perform, though.
This I can relate too. . . . . and just to add another poison, Aperture does a grand job with the A900 files, but like Guy, my poison is so incorporated into my DNA that the thought of changing it might be terminal:ROTFL:
Although - unlike Guy, I do have C1 6 and LR 3 installed on my system . . . I just don't use them!
 

Guy Mancuso

Administrator, Instructor
Michael please go ahead process some out and see what you come up with for sure. I do agree with you also on the Sensor plus issue and we are pretty much saying the same thing in a sense. I do think at some point this deserves a thread on exactly this also. Be interesting to actually test that out sometime and see how they actually match out. M9 is a good example for sure because of the Non AA filter but the MF is a much bigger sensor so that throws in a curve ball.
 

mjm6

Member
Ok, just finished...

I didn't do any tonal adjustments at all. I left the color on the 'auto' setting, and made no other tweaks.

There is a little sharpening in LR and then a little after I resized to 1200px which is what it appears you did.

The noise reduction is basically the default in LR3, with the exception that I turned n the Luminance NR to different levels depending on the ISO (20 for 800, 30 for 1600, 50 for 3200), and also set the Luminance:Contrast setting to 10 and the Luminance:Detail to 60 for 3200, 50 for 1600, and 50 for 800.
 

mjm6

Member
Here's what I see in the C1 images; a pretty tight image at 800, with a little grain and very subtle but minor low frequency chroma noise. As the speed goes up, the grain increases, but more importantly in my mind, the chroma noise jumps up considerably. Not enough to be rejecting the images, but it is present, and is most apparent in the smooth surfaces of the images.

The older rendering engine LR3 files show worse chroma and luminance noise than the C1 files in the higher speed images. At 800, it is pretty acceptable. I could tame the noise a bit more but LR3 tended to go pretty flat if you got carried away with the old rendering engine. If you look at the 3200 image, there is a certain 'crunchiness' to the image grain that I never really liked (for color work).

The New Rendering Engine LR3 files appear to have slightly better luminance noise at 800, and the difference increases as the speed increases. The chroma noise is all but eliminated in the images, with a slight bit creeping in at the 3200 setting. There is a little bit lower resolution in the image (look at the text on the shaving cream). Gone is that crunchiness completely. I could smooth out the grain a little more, but at a subtle expense to the sharpness, so that is dependent on the kind of image you want and just how 'plastic' you want the image to look.

I'm not sure that I have really made the most out of the sharpening in this quick stab. I don't know the new rendering engine too well yet, as i have been working in B&W a lot recently.

I don't mind grain, but I really don't like chroma noise. Anyone remember Kodachrome 200 and how great that looked?... Anyway, I can work with TMAX 3200 grain for an image, but absolutely detest the low frequency chroma noise that a lot of digital cameras show at higher speed. Or at least used to, as they are all getting so good now that it's really hard to complain compared to the cameras of yesteryear, or the blotchy color grain of high speed print film.

---Michael
 

Guy Mancuso

Administrator, Instructor
Looks nice need to study this more but gotta run to a shoot which the Sony is primary cam. BTW i did buy a 200mm 2.8 HS Minolta this morning from a forum member. Looking forward to that lens
 

mjm6

Member
The images are so far apart that it is impossible to compare...

I grabbed the C1 3200 image and the LR3 New 3200 image and put them together for easier comparison.



There are clearly tradeoffs for rendering, but if you are thinking high speed shooting, I think LR3 does a great job. Not that it should convince anyone to change from their preferred package, but in a pinch, it can be leveraged to do very 'good things'. Ugh, now I sound like Martha Stewart!


---Michael
 

mjm6

Member
For B&W, I often prefer to use the older rendering engine in LR3 due to the grain it produces:



This is shot at 6400 handheld at the Laurentian Library in Florence, one of the finest examples of Michelangelo's work, and an absolutely sublime place of you are into architecture.

Look into the shadow areas to see some the substantial grain texture, which works just fine for me in B&W, a lot like T-3200 used to.


---M
 

jonoslack

Active member
There are clearly tradeoffs for rendering, but if you are thinking high speed shooting, I think LR3 does a great job. Not that it should convince anyone to change from their preferred package, but in a pinch, it can be leveraged to do very 'good things'. Ugh, now I sound like Martha Stewart!


---Michael
HI Michael
Thanks for that . . . . now please could you have a go in Aperture 3 as well :ROTFL:

Mind you, I completely understand why Guy isn't even going to load LR3 - once you have a workflow and a large library, changing over to another program is really a big deal (and if you spend the months necessary to do it, then when you've finished you'll find that your old program comes out with a better version!).

I suspect that Aperture would come somewhere between the two, but I'm not sure I really want to know.

. . . . the mono chair shot is lovely.

It's chroma noise that kills a picture (IMHO of course) I don't usually mind luminance noise, and it usually looks okay on Sony shots anyway.

thanks again.

all the best
 

jonoslack

Active member
The images are so far apart that it is impossible to compare...

I grabbed the C1 3200 image and the LR3 New 3200 image and put them together for easier comparison.

I'm not really that lazy - here is a go with Aperture 3



all the best
 

mjm6

Member
HI Michael
Thanks for that . . . . now please could you have a go in Aperture 3 as well :ROTFL:

Mind you, I completely understand why Guy isn't even going to load LR3 - once you have a workflow and a large library, changing over to another program is really a big deal (and if you spend the months necessary to do it, then when you've finished you'll find that your old program comes out with a better version!).

I suspect that Aperture would come somewhere between the two, but I'm not sure I really want to know.

. . . . the mono chair shot is lovely.

It's chroma noise that kills a picture (IMHO of course) I don't usually mind luminance noise, and it usually looks okay on Sony shots anyway.

thanks again.

all the best
Jono,

Thanks for the compliment on the images. I was pleased enough with it that it made it into a calendar for my friends this year.

You are absolutely right; chroma noise is the real problem. Since we both agree, I guess that makes it true and unequivocal.

I just loaded this on my uncalibrated computer at the office, and I barely see the chroma noise in the C1 image that I can see on my graphics system at home, so this is clearly an issue of subtlety rather than real image problems.

I think this is one of the real attributes of the Sony pro cameras; they have a very neutral and natural color characteristic that does not develop chroma noise as readily as some of the other cameras. Pump up the saturation a bit, and it will probably creep up somewhat.

Regardless of the software chosen, it's remarkable that a 3200 image holds together as well as it does these days. It's a credit to the technology that images like this can be made VERY useable even under somewhat non-ideal conditions.


---M
 

jonoslack

Active member
Jono,

Thanks for the compliment on the images. I was pleased enough with it that it made it into a calendar for my friends this year.

You are absolutely right; chroma noise is the real problem. Since we both agree, I guess that makes it true and unequivocal.
:thumbs: Absolutely . . . . incidentally, I didn't do any noise reduction on the Aperture version (or anything much at all to be honest).

.

Regardless of the software chosen, it's remarkable that a 3200 image holds together as well as it does these days. It's a credit to the technology that images like this can be made VERY useable even under somewhat non-ideal conditions.

---M
I quite agree - I think also that when people compare high ISO quality between say the D3 and the A900, they will pixel peep at 100% . . . . . . and the D3 will always look better - the problem of course being that they are doing much more magnification with the Sony image.

I was really surprised recently; I had to do two wedding books that I had shot, one with a D700 and one with an A900 - I was fully expecting the low light shots from the D700 to be much better, and of course, they seemed to be when zooming in and out in Aperture . . . . but when they were in the book, the Sony pictures were much nicer, the greater resolution more than making up for the noise characteristics.
 
Top