SX510-HS & DNG4PS - Calculating color matrix won't work - RAW Shooting and Processing

Marg

15
SX510-HS

SX510-HS & DNG4PS - Calculating color matrix won't work

« on: 16 / December / 2015, 07:39:06 »

Advertisements

Hi. I've always seen that colors in DNG images vs JPEG look a bit (well, a lot) off. Only recently I've discovered that a color profile could be created by using DNG4PS, which would compare RAW and JPEG and come out with a proper color matrix.

The thing is, DNG4PS hasn't been updated since 2009 and the support for my camera is lacking. I've been able to extract some info on it by reading the resolution of RAW files (Full size, Active size) etc. which has enabled DNG4PS to read RAW files, but still it won't do color calibration. I always get a message "Poor results, can't continue". The error rates come out to be quite disturbing, such as 115%, whereas only <15% is OK.

Here are the setting that I use for Canon SX510HS.

There is probably something wrong with it which is causing the calibration to fail. Does anybody know where the correct camera parametars can be obtained?

Here is how RAW image looks in Calibration window.

For some reason it is very green. I assume some kind of error in camera parametars caused the sensor array to be misinterpreted.

*Yes, I've read the manual on how it's done. I followed all the steps. There is nothing obviously wrong with the photo of the colorchekcer that would disturb the calibration process. That's why my assumption is that the settings I've used for my camera inside of DNG4PS are not correct.

To sum it up, my questions are:
1.) Where can correct camera parametars be obtained? (for use in DNG4PS)
2.) Does anybody have an idea why color matrix calculation fails miserably?

Also, is there a possibility of using the calculated color matrix inside of Adobe Camera RAW? Actually this is a quite important question since I do all of my editing there. Can these coefficients be transfered there (--> Hue and Saturation settings for Red, Green and Blue)?

« Last Edit: 16 / December / 2015, 07:44:18 by Marg »

Logged

philmoz

3450

Re: SX510-HS & DNG4PS - Calculating color matrix won't work

« Reply #1 on: 16 / December / 2015, 15:02:46 »

The parameters for DNG4PS look ok - what does the DNG file look like in ACR?

Can you upload the DNG & JPEG to a file sharing site and post the link here.

You can also try the free 'DNG Profile Editor' program from Adobe to create camera profiles for ACR / Lightroom.

Phil.

Logged

CHDK ports:
sx30is (1.00c, 1.00h, 1.00l, 1.00n & 1.00p)
g12 (1.00c, 1.00e, 1.00f & 1.00g)
sx130is (1.01d & 1.01f)
ixus310hs (1.00a & 1.01a)
sx40hs (1.00d, 1.00g & 1.00i)
g1x (1.00e, 1.00f & 1.00g)
g5x (1.00c, 1.01a, 1.01b)
g7x2 (1.01a, 1.01b, 1.10b)

Marg

15
SX510-HS

Re: SX510-HS & DNG4PS - Calculating color matrix won't work

« Reply #2 on: 17 / December / 2015, 06:48:37 »

Quote from: philmoz on 16 / December / 2015, 15:02:46

You can also try the free 'DNG Profile Editor' program from Adobe to create camera profiles for ACR / Lightroom.

Thanks for replying.

Ok, just to straighten this one out before proceeding further:
Even if I am to succeed in creating DNG4PS color matrix, it would still be limited to DNG4PS, i.e. there is no way to transfer/convert that matrix to a ACR profile - i.e. Hue/Saturation values for each primary color? (I'm not sure what the matrix coefficients actually represent there.)
Because what I'm really interested in is creating ACR profile; I do all of my editing there. I don't have the will to do RAW-->DNG conversion everytime just to get my colors right.

Speaking of 'DNG Profile Editor', yeah I've tried it, but the thing is, it doesn't work by comparing RAW and JPEG but by taking a picture of actual colorchcker chart which should be properly illuminated etc. for the calibration to be useful, which excludes the option of just displaying it on the PC screen and taking a picture of it. In another words, I should buy an actual paper colorchecker chart and take a photo of in in two different lighting situations - daylight and tungsten? Is it so?

« Last Edit: 17 / December / 2015, 06:56:12 by Marg »

Logged

philmoz

3450

Re: SX510-HS & DNG4PS - Calculating color matrix won't work

« Reply #3 on: 17 / December / 2015, 14:47:29 »

Quote from: Marg on 17 / December / 2015, 06:48:37

Quote from: philmoz on 16 / December / 2015, 15:02:46
You can also try the free 'DNG Profile Editor' program from Adobe to create camera profiles for ACR / Lightroom.
Thanks for replying.

Ok, just to straighten this one out before proceeding further:
Even if I am to succeed in creating DNG4PS color matrix, it would still be limited to DNG4PS, i.e. there is no way to transfer/convert that matrix to a ACR profile - i.e. Hue/Saturation values for each primary color? (I'm not sure what the matrix coefficients actually represent there.)

I'm not aware of any way to do this.

Quote

Because what I'm really interested in is creating ACR profile; I do all of my editing there. I don't have the will to do RAW-->DNG conversion everytime just to get my colors right.

Speaking of 'DNG Profile Editor', yeah I've tried it, but the thing is, it doesn't work by comparing RAW and JPEG but by taking a picture of actual colorchcker chart which should be properly illuminated etc. for the calibration to be useful, which excludes the option of just displaying it on the PC screen and taking a picture of it. In another words, I should buy an actual paper colorchecker chart and take a photo of in in two different lighting situations - daylight and tungsten? Is it so?

Yes, you need a photo of a properly lit color checker, not your computer screen.

Only one photo is needed, the software will calculate both tables. Two may give a better result; but unless you have a color meter to measure the temperature and tint of the light it might be hard to get a 2600K light source.

Outside on a clear sunny day around mid morning usually works well. Keep away from strong reflections that could tint the checker (even bright clothing). Zoom in to reduce distortion from the wide end of the lens.

Take the photo as a DNG in CHDK, so long as the colors looks reasonable the profile editor will be able to handle it.
If the colors are wildly off in the CHDK DNG, then there could be something wrong with the CHDK port.

One more 'advanced' tip - if you want more control over your images in ACR, select 'Linear' on the Tone Curve as the base tone curve profile. This will remove the shadow and highlight compression that Adobe builds into their profiles. Your images will look flat and horrible when you first select your camera profile in ACR; but you will retain more shadow and highlight detail and have more information to work with in the image.

Phil.

Logged

CHDK ports:
sx30is (1.00c, 1.00h, 1.00l, 1.00n & 1.00p)
g12 (1.00c, 1.00e, 1.00f & 1.00g)
sx130is (1.01d & 1.01f)
ixus310hs (1.00a & 1.01a)
sx40hs (1.00d, 1.00g & 1.00i)
g1x (1.00e, 1.00f & 1.00g)
g5x (1.00c, 1.01a, 1.01b)
g7x2 (1.01a, 1.01b, 1.10b)

Marg

15
SX510-HS

Re: SX510-HS & DNG4PS - Calculating color matrix won't work

« Reply #4 on: 18 / December / 2015, 18:40:47 »

Thanks for advice. I somehow assumed that it is possible to transfer color matrix from DNG4PS to ACR, probably because I got confused by reading about both calibration methods at the same time. My original problem has still not resolved and to be honest, I've got no idea why. The same process seems to be working for others so I will just let go of it. In the end, what I desired to achieve is the calibration that works with ACR.

In reality, my DNG's don't look that bad, i.e. there isn't any major color cast or something that looks plain wrong, it's just that the colors are a bit oversaturated at places or have a bit wrong hue (such as street lights being rendered yellow instead of orange, or foliage that has an unusual green hue to it). I'm not all about realism in photography, but still I'd like to start from a neutral point.

What you've said about setting the base curve to linear is interesting to me because I've used to start my editing process with adjusting the tone curve in such a way that actually "undoes" the effect of the logarithmic base curve so I can can "free up" that heavy contrast/compression, otherwise I'd often resort to extreme slider values such as Shadows +100, and sometimemes realize even that isn't enough.

Thanks again for answering my question and pointing me in the right direction.

« Last Edit: 18 / December / 2015, 18:50:05 by Marg »

Logged

reyalp

14079

Re: SX510-HS & DNG4PS - Calculating color matrix won't work

« Reply #5 on: 18 / December / 2015, 20:09:12 »

Quote from: Marg on 18 / December / 2015, 18:40:47

Thanks for advice. I somehow assumed that it is possible to transfer color matrix from DNG4PS to ACR, probably because I got confused by reading about both calibration methods at the same time. My original problem has still not resolved and to be honest, I've got no idea why. The same process seems to be working for others so I will just let go of it. In the end, what I desired to achieve is the calibration that works with ACR.

It can't be used directly in ACR, but a color matrix calculated in DNG4PS can be added to the CHDK DNG code. However, as you've found out, the DNG4PS process doesn't seem to work on most modern cameras, and AFAIK no one has dug into it enough to find out why.

Logged

Don't forget what the H stands for.