New custom DNG profile for Pentax 645z

[torger]

The overall profile saturation can be controlled with one parameter when rendering the profile, so it's quite easily adjusted.

The way the profile (or rather profile software) is designed now is that first there's a normal colorimetric linear profile (no film curve). This is just the same you do for reproduction work, ie you rely on test charts and instruments rather than your eyes. The the challenge has been to construct a tone curve ("tone reproduction operator" in color science speak) which when you do an A/B switch with the flat rendering and let the eyes adapt for a second or two produces the same color appearance. That is my reference has always been the colorimetric rendering and make it appear as true to that as possible, but with the curve.

The intention of the default rendering is thus to be as "accurate" as possible, that is match a colorimetric rendering except for contrast. There are no standard models to rely on in this case (color science doesn't deal much with tone curves) so I had to develop my own model, with the help of my eyes and colleagues on the RT team. Due to that local contrast and color appearance is very much connected it's not possible to make a perfect result without using dynamic tone mapping techniques, which you can't do with a profile. In this case I (we) have chosen to let saturation be higher rather than lower in cases it's not possible to make it "correct" (since most users like that), but I guess that many in the MF audience may like a less saturated look.

For my H4D-50 I've done A/B comparisons with Phocus, and concerning saturation it's not actually any big difference. I've attached on older set of images that shows the difference between DCamProf and Phocus. A is the DCamProf neutral, B is with subjective adjustments (work in progress then, had some issues) and C is Phocus.

[Bart_van_der_Wolf]

I'm also thinking if the difference in color between P45 vs 645Z is a matter of CMOS vs CCD. Without going into the CMOS / CCD debate, I do think that there is a difference in color rendition in the two types of sensors.

Well, that is the debate at times, but it remains a myth, fiction.

I don't think CCD color is necessarily better overall, but the rendition of skin on CCD seems to be superior from my eyes.

It's the same silicon that converts photons into electrons, which then cumulates into a voltage. CMOS versus CCD is basically just a different electronic layout of connectors and added transistors. Nothing about that changes the intrinsic color sensitivity of the silicon conversion into electrons. The only thing that makes the color sensitivity of any sensor vary, is the (IR filter +) CFA transmission characteristics and the profiling, and those have nothing to do with silicon as such.

Cheers,
Bart

[Clermond]

Sorry guys for bringing confusion into the discussion by telling about my person background with the switch from a P45 to the 645z. As a matter of fact I didn't use the P45 often within the last years. On the Hassi 503CX the technology from 2005 is missing some important features as AF, fast shooting, bright VF, real wide angle lenses etc. I just sold the P45 recently and ran a demo with the new owner. It still shines in bright light and without peeping the pixels or a tenth of dynamic range the overal image is still great, the color rendering in C1 v8 is very pleasant, last not least the 20yrs old 80mm PlanarT* and the 50mm Distagon FLE are amazing sharp and have almost no CA.  *sigh*

Instead I was shooting 98% of my jobs with a 5DM3. I always liked the secret sauce of PhaseOne color science better than Adobe's. One of my clients switched to a DNG workflow (no need to discuss pro or cons) so I had always a comparison in both color renderings - and stayed for most other shots with the C1.

It's not a question of CCD vs CMOS, most of all it's a question of how the software translates the sensor read out. You find the same discussion in the film business about the pleasing ARRI colors and the videoish Sony look. An experienced operator gets great results of both, but my main aim is to get it quick and as close as possible out of camera. A camera profile would be a great step.
 Maybe it's correct by the math that skin tones get more magenta and more saturated. This is something I cannot judge, but what matters is my eye.

[torger]

Maybe it's correct by the math that skin tones get more magenta and more saturated. This is something I cannot judge, but what matters is my eye.

Yes, it's nothing wrong in that, I may have sounded defensive but I was just describing how the design process has been made and the motivation behind it. If I get a lot of time at some point I'll maybe try to involve more people in a look design and see if we can come to a broadly accepted high end look, much what Hasselblad and Phase One has been doing with their profiles. It would be interesting to do, but currently I have too much work left in the software itself.

Hasselblad looks quite different from Phase One though, so surely it will be hard to get a consensus. Personally I like Hasselblad's look design better than Phase One. I think Hassy is a bit more neutral, broadly speaking Phase One is too yellow for my taste. But that's just my taste.

[larkis]

Overall i'm liking what your profile is doing with some of my recent images I'm working on. In fact, I prefer the post processing "play" i get when your profile is applied. However, I ran into an image where some odd blocky color artifacts are visible when your profile is being used. You can private message me and I can send you the raw file that is giving me this issue. I'm using Lightroom 5.7.

[torger]

PM sent, it would be nice to have the file to debug. Thanks for testing it out.

Progress is a bit slow now though so it can take a while before I can have a look on it. I'm changing quite large aspects of the program (won't affect normal look) and I want to be finished with that before debugging, as it might disappear by itself thanks to the changes I'm working on now.

[torger]

I can't recreate the blockiness artifact on RawTherapee. I've got another user report on blocky artifact too, which I couldn't recreate either. I'll check on Lightroom 4 when I get to such a box.

Adobe does some heavy optimization in their raw pipeline, it's probably related to that. It's not the first time I've run into situations where the DNG profile works with the reference code but not in Adobe Lightroom due to their optimizations. Likely some difference in clipping behavior. Probably the profile contains something strange which robust reference code handles, but not the optimized Adobe code.

[torger]

Here's new version of the profiles;

It's generated by the upcoming next release of DCamProf v0.10.0. The software is much changed under the surface to better handle extreme range (clipping, gamut compression etc). The basic "look" stays the same though.

However there's one big visible change, (deep) blues are now rendered much lighter than before. There's both a subjective and a technical reason for this. Many Sony sensors have a sensitive blue channel with large overlap. The human eye on the other hand is not so sensitive on blue, so to render blue colors realistically dark we need to subtract a lot of blue which leads to all sorts of clipping problems. Other sensors, like a few Canons I've tested, have less sensitive blue and thus have less issues when rendering a realistic blue. Why Sony have designed the response like this I don't know, but I assume it is to improve ISO performance.

Anyway to avoid issues with "extreme blues" one cannot render normal blues realistically dark, or you would get very aggressive bends in the LUT towards the gamut edges. The solution is either to not care, or to render lighter blues. I've chosen the latter here, and the good news is that this is a subjective adjustment that many like, the reason being that lighter blues make tonality better, as we move blues into an area where the eye is more sensitive.

There are two versions of the profiles. Neutral and Neutral+. The latter is supposed to be a "better neutral than neutral", at first glance you won't see any difference, but there are some very subtle adjustments, of the same type as discussed earlier in this thread. I've fixed some artifacts that could occur in the earlier versions though.

I've had quite large interest of the Pentax 645z profiles. These ones are still generated from a CC24 shot from Imaging Resource web site. It doesn't necessarily make the profile bad (large patch sets are a bit over-rated when it comes to general-purpose profiles), but it's only single-illuminant for D50. A CC24 shot for 2856K StdA tungsten would be nice, and a D65 shot (overcast), then I could render a dual-illuminant profile which would work better as a drop-in replacement for Adobe's bundled profile. (My main focus is the software itself though, not making profiles.)

Oh, the Adobe-specific profile bug reported a few posts back should be fixed now. Didn't have time to test this last rendering (I rarely use Adobe Camera Raw / Lightroom, I do most my stuff in RawTherapee) so let me know if there are problems.

[larkis]

I will give those a try with the same image that had the previous issue. Thank you very much for doing the profiles, they are very much appreciated. If there is anything I can do to help you with more 645Z files, PM about it and I will see what I can do. I also have the 645D which has a very different sensor, but seems to be more pleasant out of the box when using adobe standard profiles.

[larkis]

So I tried both profiles on the image that was exhibiting the artifacts, and the neutral profile fixes the issue. The neutral+ version of the profile seems to keep the artifacts the same as the previous version of the profiles.

[torger]

So I tried both profiles on the image that was exhibiting the artifacts, and the neutral profile fixes the issue. The neutral+ version of the profile seems to keep the artifacts the same as the previous version of the profiles.

Thanks for testing... I should have pre-tested that but I thought it would have been fixed through the many changes in clip handling. I had only tested the neutral one. I never found out what the problem was, so I need to make a deep dive on that :-/. The problem only occurs on Adobe software (not their DNG ref code though) which I don't normally use.

I'll publish new profiles here when I (think) I've fixed the problem yet again

[torger]

It seems to be related to Adobe's noise reduction algorithm. Try turning off noise reduction and see what you get.

I'll investigate some more, but it seems like it's not the interpretation of the profile itself that is the problem, but how it interacts with the noise reduction.

I've managed to reproduce a similar but less ugly problem with RawTherapee now by increasing chroma noise reduction to maximum. Now when I can reproduce it in "my" raw converter it will be simpler for me to isolate the reason.

Meanwhile try re-adjusting noise reduction or change to the neutral profile on the problem images.

EDIT: found and solved the problem, will post update soon(ish). The reason is that some of the look tunings go all the way down to neutral, and even if very very minor it disturbs some noise reduction algorithms. I'll think I'll change the software a bit so it becomes easy to always fade out a look towards neutral to globally avoid this issue (now one need to fade out each operator, I'd done that for most, but not all, and that one caused the issue).

[torger]

Here's a fixed version. Replace the old file and restart Lightroom.

Let me know if it works.

[larkis]

All seems to be fixed now, awesome. Thank you so much, i'm really liking what the profiles are doing compared to some of the others I have tried.

[mi-fu]

Thanks again Torger! you made my 645z a far better camera now!!

[mi-fu]

Hi, I know it is a stupid question, but I couldn't googled any answer. Hope you guys can help me on this.

My question is: Now Torger made the new profiles for Pentax 645Z, which I love. I would to use those profiles for my Hasseblad CFV 50C. But in lightroom, the new profiles only pop up with the corresponding cameras, i.e. to 645Z only. How can I "convert" the 645Z profiles to Hasselblad CFV-50C compatible one?

[torger]

Hi, I know it is a stupid question, but I couldn't googled any answer. Hope you guys can help me on this.

My question is: Now Torger made the new profiles for Pentax 645Z, which I love. I would to use those profiles for my Hasseblad CFV 50C. But in lightroom, the new profiles only pop up with the corresponding cameras, i.e. to 645Z only. How can I "convert" the 645Z profiles to Hasselblad CFV-50C compatible one?

Lightroom looks for the camera name and it must exactly match. So replacing the name would be the way to do it. You can do this with dcpTool I think, or using dcamprof:

1. dcamprof dcp2json p645z.dcp profile.json
2. edit profile.json in a text editor and change name to *exactly* what Lightroom expects.
3. dcamprof json2dcp profile.json hassy.dcp

(the Pentax profile is still "beta". I have some fixes to do still, will post updates later)

[torger]

Here's updated profiles. They've been updated due to generic fixes in the DCamProf tone reproduction. The base profile remains the same, and it's probably best to keep it that way while updates are related to tone reproduction operator rather than anything Pentax 645z specific.

(The tone reproduction operator is the thing that applies contrast to the flat colorimetric "reproduction style" profile, and modulates colors so they appear accurate despite the changed contrast)

Anyway what's fixed here is a shadow over-saturation issue, most evident in deep blues. Blue has got some additional special treatment. Why blue is so special and has needed so much special care is because it's weak in luminance and CIECAM02 is weak in high saturation blues.

[mi-fu]

Lightroom looks for the camera name and it must exactly match. So replacing the name would be the way to do it. You can do this with dcpTool I think, or using dcamprof:

1. dcamprof dcp2json p645z.dcp profile.json
2. edit profile.json in a text editor and change name to *exactly* what Lightroom expects.
3. dcamprof json2dcp profile.json hassy.dcp

(the Pentax profile is still "beta". I have some fixes to do still, will post updates later)

Thanks Torger for the help! I have now successfully converted them for the use for 50c