7D, ISO:3200 sample picture

I think that this thread is heading down the same path that all 7D threads do, with people saying the noise is not acceptable to them for valid and some non valid reasons and others showing processed images, with NR, saying how good it actually is.

Others try to demonstrate that if you shoot on a full moon, standing on one leg, the camera will show (pick your defect). I am not going to comment on individual images because many of them actually do not look good, (my opinion) relative to what is possible with a digital camera, and others contain subject matter that is good at hiding noise. Then there is the downsizing.....

Not everyone shoots the same subjects under the same conditions or processes their image for the same "look" or outputs to the same media, be it web or print or publication etc.

Noisy images, from any camera, can be considered a "style" if you are so inclined. It was popular in film days with fast B&W film. People shot it, or pushed slower film, for the grain, usually, and used it to convey a mood. There was no Photoshop back then. I come from film days.

So, if someone is happy with their camera, good for them, if someone isn't, then good for them, also. People seem to take a contrary opinion as an attack on their person. It's a camera. If it, for whatever reason, does not satisfy your needs, you replace it or learn to live with it. Nobody can tell you that you should be, or not be, satisfied with a particular camera. I am assuming proper use and not an inherent characteristic of the camera. Really, that's all there is to this.

I still think one of the big issues is in the technique and tools people use to post process. I see examples constantly from LR and DPP (and other tools) where the end results are wildly different after NR.

The only thing that bothers me (granted, ive only have the camera for about a week) is the high red pixel noise in shadowed areas. I dont see this in the above photos. Does anyone else exirience this?

Not everyone processes the same way or has the same skill. Too some extent, having to process the image with some canned (same as the next guy) processing is not always desirable. If an image due to some characteristic forces you or limits your processing, I consider that a major negative.

bohdank wrote in post #12688255
Not everyone processes the same way or has the same skill. Too some extent, having to process the image with some canned (same as the next guy) processing is not always desirable. If an image due to some characteristic forces you or limits your processing, I consider that a major negative.

I have had 3 5D2s, and each one required a different level of processing. Each sensor seems to have a pretty varied way of measuring light to the point of how much noise it adds to the image. The best way is to take your equipment, study the images at different ISO ranges, and create a method or process that works best for those images from that camera at those ISOs.

Finally, depending on what your final output is going to be (web vs 8x10 print vs poster print) also comes into play, in many cases, you may not need to do anything at all, because many print sizes may provide a better image experience than pixel peeping on the screen.

I still say with certainty that whatever tools one uses to move raw to JPG and the steps they incorporate (or the default activities built into that tool during the conversion) creates a very wide diversity in final JPG output.

I have seen it too many times in these types of posts, LR vs DPP vs LR3 vs OOC JPG vs Noise Ninja vs Noiseware, etc. produce wildly different results, enough to cause many that are uneducated on this process to produce sometimes inappropriate conclusions about the hardware itself.

We also know that some of the best submissions of images on this board and others are indeed post processed by very experienced individuals, and we try to gauge our own images against their results without knowing how or what was used.

Not denying your conclusion, TS. Varying skill levels will often result in different conclusions.

I have a second 5DII that will be here Monday so I am curious how different it may be, if at all, from my current one in the aspects that count. I don't use any canned processing, by that I mean, I process by eye, so minor variations in color, contrast etc. won't affect how I do things.

bohdank wrote in post #12688631
Not denying your conclusion, TS. Varying skill levels will often result in different conclusions.

I have a second 5DII that will be here Monday so I am curious how different it may be, if at all, from my current one in the aspects that count. I don't use any canned processing, by that I mean, I process by eye, so minor variations in color, contrast etc. won't affect how I do things.

Out of 3, 2 were very close to each other, the one though was bad. It was the one that produced varying levels of pink noise in a darkroom with no light at different ISO settings. It was bad enough that I could see impacts all the way down to ISO 1600. The sensor was recording light even though there wasn't ANY. This was a closed room in the basement with no windows, no lights, and the lens cap on. Sent it to Canon, they adjusted WB and some other things, and not the root of the problem.

I wonder if there was a free tool out there that could measure "noise" on such a frame (no light dark room lens cap frame at a certain ISO)? It would be interesting to run 3-4 same bodies through that tool at different ISOs to see the variances between them.

Also, it would be interesting for someone to take the same ISO 3200 raw image, and run it through various tools to see the difference. I know we just recently had 2 different raw processed images, and you could see a difference, but maybe:

1) DPP with standard NR set on the file
2) LR1, then LR2, then LR3
3) OOC JPG with Noiseware
4) OOC JPG with Noise Ninja
etc

It would be a great way to process one raw file, then show 100% crops of the results from different methods/tools. I like those types of analytical approaches to comparing bodies.

I remember you posting samples from that defective camera.

TeamSpeed wrote in post #12688673
Out of 3, 2 were very close to each other, the one though was bad. It was the one that produced varying levels of pink noise in a darkroom with no light at different ISO settings. It was bad enough that I could see impacts all the way down to ISO 1600. The sensor was recording light even though there wasn't ANY. This was a closed room in the basement with no windows, no lights, and the lens cap on. Sent it to Canon, they adjusted WB and some other things, and not the root of the problem.

I wonder if there was a free tool out there that could measure "noise" on such a frame (no light dark room lens cap frame at a certain ISO)? It would be interesting to run 3-4 same bodies through that tool at different ISOs to see the variances between them.

Also, it would be interesting for someone to take the same ISO 3200 raw image, and run it through various tools to see the difference. I know we just recently had 2 different raw processed images, and you could see a difference, but maybe:

1) DPP with standard NR set on the file
2) LR1, then LR2, then LR3
3) OOC JPG with Noiseware
4) OOC JPG with Noise Ninja
etc

It would be a great way to process one raw file, then show 100% crops of the results from different methods/tools. I like those types of analytical approaches to comparing bodies.

Here are LR 3.4.1 and DPP3.9 compared. Shot crop with the 7D and the EF-85 f1.8 at ISO3200, f2.8, 1/40sec

Miguel

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.

paparios wrote in post #12688768
Here are LR 3.4.1 and DPP3.9 compared.

The problem with these comparisons is that they're really testing the ability of the person using them.

I own DPP, LR3, CaptureOne Express, DxO Optics Pro 6 and ACDSee. I could try processing the same image in all 5 of them and LR3 would come out on top - not necessarily because it's the best, but because it's the one I best know how to use. CaptureOne would look awful - because I've only had it a week and I've yet to try it out.

hollis_f wrote in post #12688902
The problem with these comparisons is that they're really testing the ability of the person using them.

I own DPP, LR3, CaptureOne Express, DxO Optics Pro 6 and ACDSee. I could try processing the same image in all 5 of them and LR3 would come out on top - not necessarily because it's the best, but because it's the one I best know how to use. CaptureOne would look awful - because I've only had it a week and I've yet to try it out.

Agreed, but yet, that is how all these "high ISO" comparisons are being performed (one person in one type of tool with one level of skillset vs another), which is my point. I expect that many of the differences of images I have seen posted touting the great ISO performance, or the poor ISO performance, of specific ISO values are more due to this factor than the 7Ds themselves.

I think this is a fair example above because if you just take each product and let it do its own default thing with the same raw, you can see the differences from each as a baseline, software by software. Never mind the expert knowledge one has in one tool over another person or tool to produce something better, that is yet another dimension to all of these "comparo threads".

TeamSpeed wrote in post #12688932
Agreed, but yet, that is how all these "high ISO" comparisons are being performed (one person in one type of tool with one level of skillset vs another), which is my point. I expect that many of the differences of images I have seen posted touting the great ISO performance, or the poor ISO performance, of specific ISO values are more due to this factor than the 7Ds themselves.

I think this is a fair example above because if you just take each product and let it do its own default thing with the same raw, you can see the differences from each as a baseline, software by software. Never mind the expert knowledge one has in one tool over another person or tool to produce something better, that is yet another dimension to all of these "comparo threads".

I agree. Actually, the LR3.4.1 version has more work into it. The DPP version is the default one, except for a correction on the WB. It seems that, in this case, DPP works a little better. I remember you got to similar results with your 7D.

Miguel