brainiac wrote:
erichards file is looking very familiar to me and skibum's isn't. i want a camera like skibums, not one like mine. i must try going back to 1.0.6 but i am increasingly suspicious that my camera will be going back to canon.
I doubt changing to FW 1.0.6 will improve things... I was under the same impression that the FW had something to do with it, but I also read reports of differences in IQ at lower ISO's between FW 1.0.6 5D2's. It is worth a try though... if only to rule out a negative effect of FW 1.0.7.
My guess it is hardware related. People were saying I had terrible bad luck getting three 5D2 duds... Although I agree to some extent, I think Canon screwed up big time in the early 5D2 production runs. The company has a reputation to do so...
I hope you will find a proper working 5D2 or maybe changing back to FW 1.0.6 will do the trick.
There was a similar problem with the early 5D and 20D if i remember correctly. I worry about the ability of Canon to fix it if you think you have a defective camera though so be prepared for ''camera checked and found to be within manufacturers tollerances" on your repair sheet when it's returned. It's still great to see people posting high ISO samples showing no pattern noise - keep them coming as it realy helps and informs us as to how these cameras behave at low ISO - NOT!
Well - it's good to see that I'm not going mad and am not alone, but it is also worrying because it suggests that this might be a limitation of (some of) the cameras.
Although this thread is about low ISO performance, it's still useful to see high iso examples because they show that the camera CAN do a good job in the shadows under different circumstances.
Rescuing underexposure at 100 iso with contrast set to minimum is very important because it is the best way to capture wide dynamic range without HDR. That's why this issue matters quite a lot.
Now let's see, is it the case that skibum is NOT using DPP for conversion and Richard and me are? As yet I don't think I've seen examples of this phenomenon produced by convertors other than DPP. Is there anyone out there who uses 5D2, DPP and another raw convertor, and who can show that 100 iso files at low contrast in DPP produce this mess whereas the same file in the other convertor doesn't? That would give us a reason to petition Canon for an update to DPP to fix the problem...
The 5DII file that I processed on another thread ( comparing it with the D3X) showed similar pattern noise and DPP dealt with this noise better than ACR and C1 so I don't think it's a RAW processor problem but a camera hardware problem. We still need to see some side by sides with other cameras to quantify this in relation to real world expectations.
Interesting. Let me just reiterate my complaint about this issue: I am not about to move back to an original 5D. My disappointment arises from using a 1Ds3 for a year, and Canon's claim that the 5D2 has superior image quality to the 1Ds3. At high iso that is true, but at low iso it does not seem to be.
Interesting. Let me just reiterate my complaint about this issue: I am not about to move back to an original 5D. My disappointment arises from using a 1Ds3 for a year, and Canon's claim that the 5D2 has superior image quality to the 1Ds3. At high iso that is true, but at low iso it does not seem to be.
brainiac wrote:
Interesting. Let me just reiterate my complaint about this issue: I am not about to move back to an original 5D. My disappointment arises from using a 1Ds3 for a year, and Canon's claim that the 5D2 has superior image quality to the 1Ds3. At high iso that is true, but at low iso it does not seem to be.
Not in your case anyway...
These are relevant questions IMO:
1) Is there really a difference in low and high ISO performance between different 5D2's?
2) If so, is this related to:
- RAW converter used (for example Adobe claims that LR2.2 can produce artifacts with 5D2 files)?
- FW used (either FW 1.0.6 or 1.0.7)?
- Hardware "issues"?
The only way to find out is to do some controlled side by side testing of different 5D2's (with different FW installed) and process the files in different RAW converters...
brainiac wrote:
Now let's see, is it the case that skibum is NOT using DPP for conversion and Richard and me are? As yet I don't think I've seen examples of this phenomenon produced by convertors other than DPP. Is there anyone out there who uses 5D2, DPP and another raw convertor, and who can show that 100 iso files at low contrast in DPP produce this mess whereas the same file in the other convertor doesn't? That would give us a reason to petition Canon for an update to DPP to fix the problem...
i'll try a test scene at iso 100 with dpp and cs4 when i get a chance
well it is dark now but i quickly found this ISO200 with HTP on the RAW values went over very wide range, I used like 62 highlight recovery in ACR anyway here are the (oh, also i took these on a FREAKING cold day, like I think it was about 12-14F at the time):
still not seeing that nasty pattern noise you two guys seem to be easily getting.
maybe this is not the best sample, but still seems like has to be doing better
although maybe it is not quite an extreme enough scene, since I didn't really have a huge area of truly deep shadow, the black necks are pretty skinny. I do have an ISO6400 shot under very feeble indoor lighting and there the shadows do have lots of horizontal banding (still not really with the vertical though). to really be sure need to do eact same scene setup with two different bodies.
hmm seems like perhaps DPP applies more NR even when NR sliders are set to zero, i did sharpen the acr version a bit more fiercely, but still seems like DPP did more NR by default
also don't mind the blur, those geese in the 100% crop are not within the DOF
Thanks Arne - your 1Ds3 definitely looks better in the shadows. This is exactly what I found. I also found that the 5D2 was a little brighter at the same nominal settings (i.e. more sensitive). Good to have corroboration.
Righteo, taken a quick image of my window with a bit of wall, then processed the same image in ACR, DPP, and Capture 1 Pro, applying a 2 stop push and NOTHING else. Anything that could be turned off has been, so any software noise reduction, sharpening, etc. This is as close to straight off the camera + 2stop push I can do.
I have taken a small 768 x 705px crop from the same spot just underneath the window on the left and displayed here @ 100%. Cropped in CS4 as 16bit TIFFs and exported as jpeg quality 12.
Camera settings: 5DMK2 on Firmware 1.0.6 ISO100 f/3.5 @ 1/200th sec. Everything in the custom function 'Image' section has been turned off. That includes highlight tone priority, exposure noise reduction, etc.
Images and links below incase they don't load.
The image:
adrianr.net/original_image.jpg
ACR Pushed 2 stops
adrianr.net/ACR+2.jpg
DPP pushed 2 stops
adrianr.net/DPP+2.jpg
Capture One Pro pushed 2 stops
adrianr.net/C1+2.jpg
All processors certainly create slightly different results. Make of these results what you will
No worries. I was supposed to be in London this weekend too! I was down earlier but completely forgot about the pub plan, my apologies!
I've re-taken a new image now its darker with even deeper shadows. However do everyone take the results with a bag of salt, as they've been post processed under completely unrealistic conditions. +2 stops in the raw converter, and then 50% shadows tool in photoshop. It shows the criss-crossing you mention, but then I am dragging these shadows back from near blackness.
Everything set the same as before.
Original image
adrianr.net/original_image_2.jpg
ACR
adrianr.net/ACR+2+50.jpg
DPP
adrianr.net/DPP+2+50.jpg
Capture 1 Pro
adrianr.net/C1+2+50.JPG
I do fear by this point we are just trying to find fault where there is none. Unfortunately without a 1DS3 sat next to me to compare we can't be sure!
Even under these conditions though, Capture 1 pro still produces the most pleasing result to my eyes.
