• Please use real names.

    Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed

    Firstname Lastname

    Register

    We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know!
  • Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!

Combing artifacts in 30D RAW histograms - what casues them? (Chuck Westfall?)

John Sheehy

New member
Adrian Warren said:
As a matter of interest, are the shots exhibiting combing in focus? Does the combing appear on entirely OOF shots?

I'm wondering about Airy Discs:
http://en.wikipedia.org/wiki/Airy_disc
No. You do not see the detail of the airy disks, as they are broken up by the geometry of sampling, bayer CFAs and the AA filters, and even if you could they would still be analog, both individually, and as a collection of millions of diffracted points of light.

The combing spoken of here is a simple multiplication/division with whole numbers, performed by software/firmware or circuitry; such as the numbers:

0 1 2 3 4 5 6 7 8 9

becoming:

0 1 2 4 5 6 8 9 10 12

rather than:

0.00 1.33 2.67 4.00 5.33 6.67 8 9.33 10.67 12.00
 

John Sheehy

New member
Peter Ruevski said:
Hi John,

Of course I remember you. Your early posts on the intermediate ISOs at dpreview are what started me "digging in raw data" :)
It gets worse. I am *really* starting to get very annoyed with Canon. My 20D is giving me trouble, and I wanted an EF-S backup/second camera, anyway, so I bought a 400D. First thing I did when I got home with it was to test the noise levels, and RAW saturation levels. ISO 100 and 200 had less noise relative to RAW saturation than any Canons other than the 1*mkII series. After taking a number of shots, I loaded one ISO 100 image into IRIS, and immediately I noticed that there were pixels registering 4095. That should be good, but it is not the same as the other times I looked at clipped ISO 100 data (3726, or something like that was the clipping point at ISO 100). Then, I looked at a histogram of the RAW file, and alas, the camera multiplied everything by 31/25! That's a loss of about .3 stops of highlights, because some Canon engineer thought that it might be important, for some unknown reason, to scale the original RAW capture, as opposed to simply embedding a unique whitepoint value in the RAW file's metadata.

Canon, stop screwing with *MY* negatives. Now I can't even assume I know how much DR the camera has, as it may throw away .3 stops of highlights (and DR) with a seeming roll of dice, forcing me to expose lower and get more noise.

What in the world is Canon thinking with this nonsense? Canon, please leave the RAW data alone. The way it works in a 20D is a beautiful thing. 128 for blackpoint at every ISO, and values up to 4095 for every pixel. The 30D introduced inter-math based extra ISOs - yuck. It also scales the RAW data with integer math on occasion for the base ISOs as well, mildly in the 30D, and now sickly in the 400D.

Leave it alone, Canon. Just put a different whitepoint in the metadata, to instruct the converters to scale values. LEAVE THE RAW NUMBERS ALONE.

You thought your 27/26 scaling was bad!
 
I will get burned/flamed for this......

I know it's always better to do something else, but why not just do what the camera is good at, taking pictures.

I'm a perfectionist when it comes to my photos but I look at the endresult and work with that, in other words, I learn to use the tool which is the camera.

The noise levels on the 5D are so wickedly low that even prints on A4 format from ISO3200 are almost clean, to be honest I couldn't care too much for numbers, the end result is what counts.

Highlights are a problem with digital, that is known for a long time, fuji is better with highlights thanks to the design of the censor, also that is known, but when I look at the sharpness and depth/color of the endresults I want my Canon :D
 

John Sheehy

New member
Frank Doorhof said:
I'm a perfectionist when it comes to my photos but I look at the endresult and work with that, in other words, I learn to use the tool which is the camera.
The digital negative is mine. It should just be a digitization of the amplified sensor readout. Canon is compromising the integrity of my negative for no good reason. There is absolutely no need to scale RAW data after digitizing it; if the camera thinks that the amplifier didn't achieve the target amplification, it should just put a metadata in the RAW file that says that the data needs to be scaled by 1.1923 (31/25), or state that the whitepoint is 0.8387x (25/31) what it normally is. What they do, instead, is make the data less accurate and throw away dynamic range. I don't see any value in being complacent about that.

One reason I didn't even consider the 30D was the make-believe ISOs it has, and the fear of accidentally using them, or having to dial through them, to get to the ones I want. I thought that since the 400D didn't have these extra ISOs, it would be straight-forward with the RAW data, avoiding the nonsense.

The noise levels on the 5D are so wickedly low that even prints on A4 format from ISO3200are almost clean, to be honest I couldn't care too much for numbers, the end result is what counts.
Like it or not, your image is recorded and developed as numbers. Better care should be taken of those numbers, IMO.

The noise levels on the 5D are nothing special, compared to other Canons at the pixel level. The main benefit is that there are 12MP, and that allows more resolution, and the fairly large pixels each have potentially better contrast with their neighbors, because the pixel pitch is less demanding on the MTF of the optics. The 5D exhibits posterization gaps in its data, too; at least some copies. It is inexcusable; it is just a horrible way to deal with what may or may not be a problem, however subtle. It is not a sacrifice for something better; it is just needless destruction. The person who made this decision is not a good steward of data. There are better ways to deal with the issues, real or imagined.

Highlights are a problem with digital, that is known for a long time,
This is a different problem. The dynamic range is *changing* from frame to frame at ISO 100 on my 400D, by 0.3 stops. That means that assuming that the higher range is available will be risky when the occasional lower-DR frame happens.

fuji is better with highlights thanks to the design of the censor, also that is known, but when I look at the sharpness and depth/color of the endresults I want my Canon :D
The Fuji method of increasing DR does nothing to improve sensitivity; it simply de-sensitizes half of the pixels so they don't blow as easily. The "shadows" of the less-sensitive sensels add noise if they are used for increasing resolution of the overlapping ranges. I haven't seen the Fuji software, but if I were writing it, I would provide a slider that acted like a crossover frequency in audio, only as a crossover point for the light levels to fade the two images for the overlapping areas. It might be better to leave out the extra resolution if it increases noise.

I want my future increases in DR to come from the shadow end of things. I already have ND filters.
 

Peter Ruevski

New member
A lo-o-o-o-o-ong philosophical post

Frank Doorhof said:
I will get burned/flamed for this......
Not at all - in fact your post made me think why some people (including myself) pay so much attention to the technical details of photography sometimes. After thinking about it a bit I think I can answer for myself:

Frank Doorhof said:
I know it's always better to do something else, but why not just do what the camera is good at, taking pictures.
This is the crucial question. And in a long and indirect way it leads to the answer of my own question above.
Taking pictures – for me - has always been easy, pleasant and a lot of fun – I have been doing it since I was six. Buy a roll of film, shoot, take it to the lab, get the (black and white) prints, enjoy - nice and simple.
Then in high school I started to develop and print my own black and white photos. Initially I was learning and was happy to just get a decent looking recognizable photo. But as I learned I started to see that a photo is not an "absolute truth" – something which strangely enough I had not realized until then – and I was quite "shocked". I started playing with different contrast papers and developers, burning, dodging and… something for which I do not know the proper term in English – when you expose the paper to white light for a short time half way through development. I would spend inordinate amounts of time (nights) and paper printing the same photo again and again trying different things.
So now the look of a print had become to a large extent a matter of my interpretation and while I found that slightly disconcerting I actually enjoyed it.
I certainly spent much more time in the dark kitchen at night then I spent shooting :)

Then color film became cheap enough (I am not a hundred years old – I simply lived in Eastern Europe:) and I started shooting only color. The dark room equipment started collecting dust in the wardrobe and photography was once again an absolute thing for me. Buy a roll, shoot, take it to the (cheapest possible) lab, get prints... there was no accessible/cheap pro lab and therefore there was no custom printing or "I do not like these prints" for me. The operator of the minilab made all the decisions for me and in fact I kind of liked it, my mother certainly liked it – the kitchen was once again safe at night :)

... a few years pass ...

Which brings me to May of this year when I eventually bought a 30D. Being a programmer I did not even for a second consider shooting JPEG – it is an ATROCIOUS destruction of data - and here I totally agree with John! I did shoot RAW+JPEG though and for a whole month did not touch the RAW files. But then of course I went back to the "darkroom" or rather lightroom (is this trademarked by Adobe already?) and I have to say having color – at least for me – makes it much harder. Now the look of my photos is even more a matter of my interpretation – and I still have not gotten used to it.

To finally answer your question "why not just do what the camera is good at, taking pictures" - I am, and since that is as easy and pleasant as it has always been I do not need to discuss it in forums :) The artistic side of developing RAW files on the other hand is still too hard for me and I do not consider myself qualified to discuss it – for now I am still reading and learning.

But treating RAW data as numbers (and it is numbers!) and analyzing it in every possible way - I am very qualified to do - and feel confident to comment on what I find. Therefore when I found something interesting I decided to share it.

As to why the combing is bad I think John answered very-very well in his last two messages.
 
I understand :D
And I love in depth reviews and discussions but I just worked my way to some threads on DPR where people are bashing cameras on technical things only, which normally is no problem at all.

That's why I wanted another view :D
 
Top