New custom DNG profile for Pentax 645z

[torger]

As the off-topic-guy I am I posted a DNG profile for the Pentax 645z in this other thread http://forum.luminous-landscape.com/index.php?topic=104101.0 To gain a bit better visibility for Pentax 645z users I post here in a new topic.

I don't own a Pentax 645z but I've seen some interest in improving on the default ACR profile, so I just made a quick DCamProf profile for demonstration. I think it's unfortunate that the Pentax with the great Sony MF CMOS sensor shouldn't have good profiles like other MF offers with the same sensor (IQ250, Credo 50).

As before the profiles attached here are made from this online CC24 shot: http://www.imaging-resource.com/PRODS/pentax-645z/645ZhVFAI000100.DNG.HTM . That shot is made in artificial daylight and suffers a bit more glare than a really high quality shot colorchecker shot would do, but I think it still did well as basis for a D50 profile. It would be great to have a Tungsten (StdA) shot too to make a dual-illuminant profile like the bundled is, but then someone needs to provide a shot. Tungsten+D65 is the standard combo for dual-illuminant, but StdA+D50 is fine too.

I've re-rendered the profiles from the original thread and added the ColorMatrix from Adobe's bundled profile. The reason for this is to avoid a white balance shift (caused by an unfortunate design decision by Adobe), discussed here: http://www.ludd.ltu.se/~torger/photography/camera-profiling.html#white_balance_shift it won't affect color rendition, as a different matrix is used for that (ForwardMatrix). For the same reason this is a "fake dual-illuminant" profile, it says it's a StdA+D65 profile but it's in actuality only D50, this to report the exact same color temperature as the bundled profile.

The curve in the profile is Adobe's default tone curve, just as the bundled profile. I don't have Pentax's own Digital Camera Utility 5 which I otherwise would check if they have some different contrast there and match the curve with that, it's probably a better reference than Adobe, but possibly it's the same result.

The DNG files I've got have embedded JPEG in them, usually one can look at them to get the camera's own rendering. That rendering is simply the default Adobe curve with the embedded matrix (no LUT), ie very simplistic rendering. I'm surprised if that corresponds to in-camera JPEGs. Maybe the embedded preview in PEFs looks like the DCU5 output? Anyway, with the DNG files and no DCU5 software there's no ability to try relating to the camera's native rendering, unless that really happens to be matrix+acr default curve (which I doubt).

Suggested further work would be:

• Get StdA/flash/D65 shots CC24 to make profiles for other light conditions / DNG profiles
• Relate to the native DCU5 rendering, mainly the tone curve, often it's preferable to mimic at least some aspects of the in-camera/native rendering
• Fine-tune the look
• High saturation blue may have a clipping issue (I've seen it in some test pictures), can be adjusted

I have attached one with an example subtle look adjustment, as described here: http://www.ludd.ltu.se/~torger/photography/files/custom-look.json
This look is a generic one, not made specifically for the Pentax. Looks does not necessarily need to be made specifically per camera, but for example a very subtle skin-tone hue change would likely be camera specific, while changes like warmup of green/yellow midtones are more generic.

The difference between DCamProf and software like X-rite is that quite much effort has been put into rendering colors well after the tone curve has been applied, while most (all?) commercial profile makers make colorimetric base profiles and just slam a tone curve on top and let it become what it becomes. A lengthy discussion about this can be found here: http://www.ludd.ltu.se/~torger/dcamprof.html#tone_curves

DCamProf is probably not the best profile maker if you're going to make a reproduction profile (no tone curve), then the commercially available alternatives are probably better. But for general-purpose photography it does very well. The intention is to be able to match and exceed what the best bundled profiles provide (the bundled profiles are made with in-house software not available to consumers).

While the look can be fine-tuned, and small improvements of color checker shots can be made, if you don't like the look at all of these two profiles, then DCamProf profiles are not for you. These profiles are very close to what a "final" profile would look like, at least in lights close to D50. I would myself not doubt using them in production, especially after looking at the result of Adobe's bundled profile

[mi-fu]

Just tried it. The results are very impressive! Thanks a lot!!!

Actually, Torger, do you happen to have the profiles for H4D-40? Because I always use Lightroom for the sake of its multiple functionalities, but the default adobe color is just far behind Phocus. It would be great if you can share the profile for Hasselblad users.. 

[torger]

I have a Hasselblad H4D-50 myself for which I have hand-tuned a profile and I'm becoming pretty pleased with that, although testing is mostly for my own stuff I like to shoot (landscape). I haven't really assembled it though, maybe I can do it tonight. Both a neutral profile, and one with the hand-tuned subtle look. I doesn't look exactly like Phocus, but not so far from it. I haven't intended to copy Phocus, but we come close anyway due to that Hassy also likes neutral realistic renderings.

I think the Hasselblad H4D-40, H4D-50, H5D-40 and H5D-50 all use Kodak sensors with the same color filters so they should be compatible, that is my H4D-50 profile should work with H4D-40 too.

[mi-fu]

Just to post a quick pic to illustrate the color difference. Above is using Adobe default color profile. Below is using the one Torger provided (dcamprof-v0912-pentax-645z-look.dcp). I didn't change the white balance. I just leave it as "auto" by the camera.

[mi-fu]

I have a Hasselblad H4D-50 myself for which I have hand-tuned a profile and I'm becoming pretty pleased with that, although testing is mostly for my own stuff I like to shoot (landscape). I haven't really assembled it though, maybe I can do it tonight. Both a neutral profile, and one with the hand-tuned subtle look. I doesn't look exactly like Phocus, but not so far from it. I haven't intended to copy Phocus, but we come close anyway due to that Hassy also likes neutral realistic renderings.

I think the Hasselblad H4D-40, H4D-50, H5D-40 and H5D-50 all use Kodak sensors with the same color filters so they should be compatible, that is my H4D-50 profile should work with H4D-40 too.

If you can share them, it will be great!! I think many Hasselblad users will be very happy.

Actually, I'm also wondering if the profiles for Pentax 645Z will yield some good results with Hasselblad 50C backs, as both of them are using the same sensor. Or the in-camera algorithm will lead to very different results?

To me, color profiles have always been the biggest issue when using Lightroom. And I really don't enjoy switching among LR, C1, and Phocus. It is such a pain. I looked at the DCamProf introduction you linked. Now I'm tempted to make the profiles for myself, though I think there is quite a bit of learning curve.

[torger]

I haven't tried them on the H5D-50c, but probably the look will be very similar. I don't think they differ much in response.

Worth mentioning is the Credo 50 which is a special case though as it uses a different ICC pipeline than the Phase One backs, making it impossible to use IQ250 profile on Credo 50 back despite the same sensor. This has lead some to believe that the hardware response differs a lot, but I doubt there is much of a difference there either, but Leaf had their own ICC pipeline in their raw converter and that was incorporated to C1 when Phase One bough Leaf.

Should be said that DCamProf is a work in progress. It's at version 0.9.12 now, and when it reaches 1.0.0 that should mean it's ready for "broad" use. Today it's early adoption, and since I've seen a sharp rise in use I also get more question and also some issues.

Today I discovered I need to rework the gamut mapping algorithm. This means that DCamProf is currently a bit weak when it comes to rendering ultra-high saturation colors, so you may discover some issues there also in the posted 645z profiles. I can post updates later.

I'll focus work on fixing bugs and reaching 1.0.0, so it can take a bit more time while getting those Hassy profiles put together...

[mi-fu]

Thanks Torger for your work!!! 

[torger]

Here's updated profiles after I've fixed a clipping bug and also improved gamut compression algorithm (the look has gamut compression, the neutral has not).

[ManfredS]

Hi,
I would like to use the profiles, but I can somehow not install them. I am running the latest version of Lightroom on Mac OS. Can please somebody help.

Thanks

[jwlimages]

ManfredS wrote:

I would like to use the profiles, but I can somehow not install them.

-- well, if you're on a Mac, copy them into YourUserFolder/Library/Application Support/Adobe/Camera Raw/Camera Profiles

(when you download them, if they don't automatically expand from .zip files to .dng files, just double-click on them, then install)

If you're on Windows, someone else will have to chime in.

Good luck,

John

[jwlimages]

Just wanted to add my voice to JCL212's:

Thank you, torger, very much for your good work!

John

[Rolek]

I've just tried these on a few recent landscape photos and compared the results to the Huelight profiles.
There is a big difference in how these render the shadows - in a much more positive way!
Where I lifted my shadow slider up by a value of 29 I could reset the slider completely and add some more contrast, because otherwise the look is quite flat...which I like better as a starting point however.
Greens have more punch due to a perceptional shift (which may not me what your eyes might be seeing, because I have a red/green weakness) towards yellow. In direct comparison, the profiles made sunlit grass and greens in general appear vivid and real. Unfortunately all the pics I've tried so far are of sunlit test shots of nature. And I love the results  from the "look" profile especially.
Thanks alot for your work and for sharing this with us!!

[torger]

Nice to hear.

I'm going to rename "look" to "neutral+" or something when I release something more, because the look is so subtle it's not really intended to throw off neutral in any major way, just tune things that just seems to work better than the plain neutral.

I will release some update later, it won't change the look in any major way, but I'm still working on extreme saturaton color handling where DCamProf falls a bit short. If you shoot night scapes with lots of artificial lights it's likely you'll see some issues around the lights. In these color ranges the challenge is that the colors are so strong and narrow band that the equations become unstable and thrown into undefined space and all sorts of bad things happen. It's most easily seen in the blue range. For normal landscape scenes you won't get into that color range though so there it should work fine.

[alatreille]

Anders,

Thank you for this work!
Most appreciated.

Cheers

Andrew

[lenticularis]

I have a Hasselblad H4D-50 myself for which I have hand-tuned a profile and I'm becoming pretty pleased with that, although testing is mostly for my own stuff I like to shoot (landscape). I haven't really assembled it though, maybe I can do it tonight. Both a neutral profile, and one with the hand-tuned subtle look. I doesn't look exactly like Phocus, but not so far from it. I haven't intended to copy Phocus, but we come close anyway due to that Hassy also likes neutral realistic renderings.

I think the Hasselblad H4D-40, H4D-50, H5D-40 and H5D-50 all use Kodak sensors with the same color filters so they should be compatible, that is my H4D-50 profile should work with H4D-40 too.

Thanks for this good work!

I do not have a P645Z however I do have a 645D . I believe that the H4D-40 sensor may be the same Kodak chip/sensor as the 645D. Are you able to conform that or not please? If so, noting your comment above, have you shared your H4D-50 profile already?

many thanks,

David

[torger]

Yes if we're lucky the profile would be compatible with the 645D too.

No I haven't released it yet. I'm working with the profiling software to fix some issues. I need to do some actually paid work too inbetween which is slowing progress . While there are issues in the software I'm not releasing as many profiles as I will need to regenerate them anyway. I'm thinking of scripting it so I can have a web page of example profiles that I can automatically re-render when the software is updated, but I haven't come that far on that yet.

For the impatient a workflow to make your own profile is here (it's basically just a copy and paste of commands so it's easier than it looks):
http://www.ludd.ltu.se/~torger/photography/camera-profiling.html#the_easy_way
Software builds for mac and windows can usually be found in this thread:
http://forum.luminous-landscape.com/index.php?topic=100015.0 (be sure to check from the back)

[Clermond]

thanks for your work and thanks for sharing. I must say at skin tones I don't see an improvement. Switching from a PhaseOne P45 to the 645z the skin tones ooc in the CaptureOne Workflow are the only thing I miss. More subtle, brighter and less magenta. The profiles both push the saturation in skin tone. For my taste the profiles go into the wrong direction. ymmv

[torger]

thanks for your work and thanks for sharing. I must say at skin tones I don't see an improvement. Switching from a PhaseOne P45 to the 645z the skin tones ooc in the CaptureOne Workflow are the only thing I miss. More subtle, brighter and less magenta. The profiles both push the saturation in skin tone. For my taste the profiles go into the wrong direction. ymmv

Less magenta means more green, so it's delicate balance :-). The provided profiles contains no skin tone hue adjustments. As it's much a matter of taste the default profiles are kept as realistic as possible, tuning it to my taste would be an exercise in frustration as you can't please everyone. Depending on light of the test shot and LUT bending etc the actual hue can differ 1-2 DE from a fixed reference.

I've noted that many profiles for higher end cameras render tones desaturated. I don't think my profiles push saturation, but rather the P45+ is rendering tones a bit low on saturation. For post-processing artists this is not really a disadvantage, it may be the other way around as it may be easier to push saturation than decreasing it.

It should be noted that DCamProf allows manual adjustment of all those aspects though for those that want to fine tune their results to their liking. For a $10k+ investment in camera gear it may actually be worth it.

However, although adjustable you can't really copy a look as that requires the profiling software be designed exactly in the same way as the profile software used for the profily you try to copy. DCamProf relies quite heavily on CIECAM02 for it's adjustment engine, while I'm quite sure that P1 and Hassy is more RGB-based.

[mi-fu]

thanks for your work and thanks for sharing. I must say at skin tones I don't see an improvement. Switching from a PhaseOne P45 to the 645z the skin tones ooc in the CaptureOne Workflow are the only thing I miss. More subtle, brighter and less magenta. The profiles both push the saturation in skin tone. For my taste the profiles go into the wrong direction. ymmv

Actually I think the skin color with Torger's profiles is still clearly better than the default Adobe profiles (more accurate red & orange). Nonetheless, I do agree that it looks more saturated than CaptureOne workflow and Phocus workflow (I also use a P30 & H4D-40, as well as 50C). But the real "problematic" area is saturated blue, as mentioned by Torger already.

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

But I'm not really an expert on this. Looking forward to hearing other opinions

[torger]

I haven't seen any technical explanation on why CCD would be better than CMOS due to the different technologies, so my best guess is that the difference in color that some see is due to something else than that it's CCD. If there existed two cameras with the same color filters on the sensor but one CCD and one CMOS and shoot a series of portraits with those and using the same color profiles I think the results would be the same, except for pixel peeping when noise would be visible. Unfortunately I don't think there are any such cameras, so the experiment can't be made.

Anyway, to me that's a bit like discussing which way to turn the loudspeaker cables in a hifi system. Now when MFD has CMOS too there's no fanboy need to favor one or another either so it has become even less interesting I think. I'm more interested in the aspects that have large effect and that you can control, and camera profiles have very large effect and at the same time you can make very subtle fine adjustments.