TuringTest
Member
- Messages
- 42
- Reaction score
- 9
Do you agree that shooting at ISO 100 disables the AFE? If yes, then what is the problem? You can always set the ISO = 100 when you shoot if you dislike the AFE.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Are we too critical with the Foveon sensor????
- Thread starter motomanDK
- Start date
-
- Tags
- nikon d750 sigma sd15
Sorry, I was replying to his statement, not your question.
I agree.
Roland Karlsson
Forum Pro
Yes, you can use ISO 100 and then adjust the exposure for under exposure instead. That gives the same result.
But - it is not as convenient. Partly because the image on the LCD will be black. Partly because the preview image in SPP will be black. Moreover, for some obscure reason, Sigma has chosen to limit the exposure compensation in SPP to +-3 stops. So, you can only use it up to ISO 800.
Roland Karlsson
Forum Pro
TuringTest
Member
- Messages
- 42
- Reaction score
- 9
And the in-camera JPEG will be dark. But you can make it work.
TuringTest
Member
- Messages
- 42
- Reaction score
- 9
I'm not Ronald, but why would it be wrong? What is so confusing about it? The raw image depends solely on exposure = amount of light hitting the sensor. The pixel doesn't 'care' what ISO you set on the dial.
TT, I am not Roland but the AFE is never "disabled" (wrong word). It it is always functional.
The AFE gain at ISO 100 is not known to the user, TT. It is not necessarily unity, if that's what you're thinking. Also, different cameras have different 'base' ISOs, e.g. 200 ISO for some Nikons, which makes your suggestion unnecessarily restrictive.
Agreed, Roland.
Even worse: for my cameras, the limit is +/- 2 stops in SPP 3, if the 'help' info is to be believed.
Scott, that is what I understand Roland meant with reference to ISO-less cameras like your SD14, and, therefore, what you have said is indeed correct.
I have proved it using RawDigger. A shot of a scene taken at various ISO settings shows identical raw images and histograms in RawDigger - but different brightnesses in SPP. Then, if you take those same X3F files and use a utility to change the ISO meta-tag that SPP uses, they can all be made to look identical in SPP. Not that I am suggesting any such tests or manipulation as a workflow.
Scott, just making a statement like that leaves an air of uncertainty in the thread.
Why does it seem wrong to you?
--
Cheers,
Ted
Last edited:
A picture is worth a thousand posts . . .
Numbers are typical.
It tries to show what happens for two ISO settings, 100 and 400, and two exposures, 0.8 lux-sec and 0.2 lux-sec. the X in the AFE block represents a PGA (Programmable Gain Amplifier). It is an analog amp, i.e. voltage goes in and voltage comes out. It's control input (ISO) is digital from the camera firmware. As can be seen in the AFE system, the PGA corrects the 0.2 exposure by multiplying the sensor output by 4. Thus the PGA, and therefore the ADC output, is the same for both exposures.
This same function in the ISO-less system is done digitally in SPP when the raw data is converted to the working image file.
--
Cheers,
Ted
Numbers are typical.
It tries to show what happens for two ISO settings, 100 and 400, and two exposures, 0.8 lux-sec and 0.2 lux-sec. the X in the AFE block represents a PGA (Programmable Gain Amplifier). It is an analog amp, i.e. voltage goes in and voltage comes out. It's control input (ISO) is digital from the camera firmware. As can be seen in the AFE system, the PGA corrects the 0.2 exposure by multiplying the sensor output by 4. Thus the PGA, and therefore the ADC output, is the same for both exposures.
This same function in the ISO-less system is done digitally in SPP when the raw data is converted to the working image file.
--
Cheers,
Ted
Last edited:
TuringTest
Member
- Messages
- 42
- Reaction score
- 9
A picture is worth a thousand posts . . .
Numbers are typical.
Very nice. What's your view on the importance of the analog gain for image quality?
IMO it is extremely important. For example, given a 12bit ADC, at ISO 3200, without analog gain, I will only get 7 bits of data per pixel ( 2^12/32 = 2^7). This is extreme quantization, and it will make the job of noise reduction much harder.
The drawback of clipped highlights in low light is comparatively minor.
Not wriggling on the hook but, if the question refers to ISO-less vs. AFE cameras, the subject has been beaten to death elsewhere and usually involves intense discussion on this kind of noise versus that kind of noise versus Poisson statistics plus what Nikon does versus Canon etc., until one's head hurts.Very nice. What's your view on the importance of the analog gain for image quality?A picture is worth a thousand posts . . .
Numbers are typical.
Yes, that explains to an extent why early Sigma cameras had limited ISO settings: 400 on my SD9, 1600 on my SD10 and SD14. 1600 ISO gets me 256 levels on those cameras. Your equation above doesn't work by the way - 3200 ISO gets you 128 levels, not 7.
Generally true, FWIW.
--
Cheers,
Ted
Last edited:
TuringTest
Member
- Messages
- 42
- Reaction score
- 9
It actually does. It's 7 BITS, not levels. With 7 bits you get 128 levels: 2^7 = 128.
My mistake. I'm not used to using 'no of bits' in that way, but it's valid enough I suppose.
So, back the earlier point:
Now I'm wondering why 7 bits is considered "extreme" when 8 bits is all you get in a JPEG image or on your monitor screen. Would you be so kind as to explain what you meant by "extreme quantization"?
Last edited:
TuringTest
Member
- Messages
- 42
- Reaction score
- 9
Those 8 bits in jpegs are after all image processing operations/calculations. For best results, you want to do your math on 12 bit data, and goto 8 bits at the very end.
Now this is theoretically correct. The proof is in the pudding: is the Quattro (which has an analogue amplifier) much better at ISO 3200 than the Merills, which do not? Ought to be.
OK, but what did you mean by "extreme quantization"? I'm having difficulty visualizing "extreme" for some reason, sorry.
I own neither model, so couldn't say. We would, however, be comparing 14-bit with 12-bit raw files, if that is relevant.
--
Cheers,
Ted
Last edited:
TuringTest
Member
- Messages
- 42
- Reaction score
- 9
Going from 12 bits to 7 bits is extreme.
Spatial resampling does not necessarily affect the bit-depth of the image.
(Imagine that the image changes gradually from a value of 0 (on the left side) to a value of 2^12-1 on the right border. If you do your spatial resampling you'll end up with the same range of values in the final image).
What you need to do is quantize the values of the pixels to 128 levels. You'll end up with horrible staircasing in the sky, etc. All smooth transitions will affected.
14 bits? Wow. That's even better.
OK
Yes, I realized my error . . . as soon as the DPR editing timer ran out. Knew you'd spot it anyway.
The easy way is to use Levels in PhotoShop. I've just made a '7-bit' image which has every other level empty thereby by making it effectively 7-bit. Doesn't look much different to 8-bit:
view original size for best comparison.
--
Cheers,
Ted
Last edited:
I'm not sure that a low exposure represents "going from" 12 bits which implies that something started at 12 bits tone depth and was somehow transformed down to 7 bits. In an ISO-less camera, this is not the case. The image simply occupies the lower seven bits of the ADC.
In SPP, or in any converter with auto-brightening, those 128 levels are stretched to effectively 256 levels (8 bit) for the display or for any JPEG conversion. I have made such an image and it looks fine on my monitor with no sign of stair-casing or posterization in the sky. See for yourself:
Please view original size to see the histogram.
The method was simple. In Levels, reduce the highest level to 127 and save the now-dark image. Re-open and brighten back up to the original brightness. Alternate levels are now missing from the image making it effectively 7 bit.
If you disagree with this, I challenge you to post an image that confirms your statement re. jaggies and staircasing.
--
Cheers,
Ted
Sorry, you didn't mention jaggies which of course are to do with spatial resolution, not tone resolution.
Roger
Senior Member
Hello AgainI was reading about the new Nikon D750 and watched some sample pictures - i saw this one where 2500ISO was used (100%):
I know that it is 2500ISO, but still this is claimed to be a high ISO camera - i remembered that i took some shots yesterday in the Disney Store with 400ISO. I KNOW there is a big difference between 400 and 2500 ISO, but still ........
Here is part of my 400ISO Disney shot (100%):
(The purple glow is due to purple light in the figure setup - NOT Foveon error!! :-D
Another example with SPP settings - please note that both pictures are with my SD15 - known for noisy pictures!
Any comments?
--
/Henrik - Denmark
(SD14 and 2 * SD15 ... and about 5 kg glass)
We are way to negative about Sigma and the Foveon. I include myself in this statement. Many a time I've said not such nice things about Sigma and the Foveon but I keep using them and buying the gear because I do realize a lot the issues I have are not the fault of the Sigma's or the Foveon but mine. I do admit that the Sigma bodies are not as good as they could be and are over priced. I've said before and I'll say it again you have to shoot the Foveon like you shoot slide film. In the film days I had 4 camera bodies and 4 lenses, four types of film, B&W ASA 125 ASA 400 Color Slide film ASA 100 ASA 400. Now I have the DP's and SD's that do the same thing or close to it. Although I do still shoot B&W film.
So before you or any one else dogs the Foveon you may want take a good look in the mirror first and ask which is the real problem, you or the camera???? Your right the Foveon doesn't shoot high ASA the Sigma's AF is spotty but it does have a HP 100% viewfinder. I shoot against guys who have Nikon, Canon's, and Leica's and they all admit that the Foveon has the best IQ of all bar none.
I have a friend who is shooting a loaner Nikon D750 right now and I can tell you for a fact that my SD15 IQ is as good if not better. My SD1 with the right lens kicks it rear end. Ok, I wish I had the AF speed and processing of the Nikon with the weather sealing of the Olympus. I can get loaner gear from different companies as my friends do but I buy Sigma for the Foveon. As for SPP I find it works fine it's like having 2 programs in one. A color and B&W program rolled into one. The best of both worlds for free, what's not to like?
Ok as for the DP's being the Poor Man's Leica? Well they are. Their IQ is as good their lenses are as good and the Foveon is better. There again the weak link is the body but I must say the DPQ's are way better than the older bodies. I really hope Sigma builds a SDQ or something that will take my SA lenses even if I have to use an adapter like the Olympus does. You have to shoot the DP's like a Leica one body one lens. You'll need to add a viewfinder but outside of that what's the difference? Ok you can change lenses but the DP's lenses are match to the sensor so would you rather have a Leica with 3 lenses or 3 DP's? Pick your poison. What's your wallet say?
If there is one thing that does concerns me it's that Sigma may stop making cameras no matter what they say this could become a reality in today's camera market leaving us with the choice, Film or Bayer?
That's it
Have fun
Roger J.
Similar threads
- User review
Keyboard shortcuts
- f
- Forum