Colour Profile Problems Part 2
Hello all
Hopefully Drew will allow this thread to develop.
The 7D2 colour cast problem is blindingly obvious to me. I have the same problem with Cap1 and 7D.
In order to see it, just process and compare on the same screen. It is differences we are looking at, not absolutes.
Cap1 is not alone in this cast problem. I suspect many raw converters have been tweaked in the last year or so to make outputs look 'good'. But not realistic, I suggest.
I am fed up with fiddling around with colours in the yellow-orange-red-magnolia hue range (i.e. autumn leaf colours). I have now gone back to the camera manufacturers' own software for basic raw conversion. DPP and IDC do a good job to my eyes and on my screen. Then I process the basic tif outputs in 'proper' software, including Cap1.
In the past my main criterion for a good raw converter was noise control. Now, with modern cameras and refined software algorithms, noise is less of a problem and I am focused on colour tones. The least offensive ones (IMO) are now DxO-10 and ACDSee-8. Perhaps by next year it will be Cap1 v9?
Always ready to learn!
Peter
Hopefully Drew will allow this thread to develop.
The 7D2 colour cast problem is blindingly obvious to me. I have the same problem with Cap1 and 7D.
In order to see it, just process and compare on the same screen. It is differences we are looking at, not absolutes.
Cap1 is not alone in this cast problem. I suspect many raw converters have been tweaked in the last year or so to make outputs look 'good'. But not realistic, I suggest.
I am fed up with fiddling around with colours in the yellow-orange-red-magnolia hue range (i.e. autumn leaf colours). I have now gone back to the camera manufacturers' own software for basic raw conversion. DPP and IDC do a good job to my eyes and on my screen. Then I process the basic tif outputs in 'proper' software, including Cap1.
In the past my main criterion for a good raw converter was noise control. Now, with modern cameras and refined software algorithms, noise is less of a problem and I am focused on colour tones. The least offensive ones (IMO) are now DxO-10 and ACDSee-8. Perhaps by next year it will be Cap1 v9?
Always ready to learn!
Peter
0
-
Radcliff camera with the CO1v8 Sony A7r standard profile. Pleasing yet absolutely spot-on. Just to show that there is no reason why people should be forced to get whacky colors out of CO1.
With profiles like these CO1 is a very solid raw converter. And this b.t.w. is not just in bright sunny daylight. In all sorts of circumstances the profile just gets it right.
Chris
http://www.vogelsenhunleefwereld.nl/A7r_DSC0384.jpg0 -
This thread is developing a life of its own. Thank you for your latest contributions.
I have previously agreed with Keith Reader that there seems to be some kind of systemic colour problem with v8.1 (compared with v7). But we don't seem to be any nearer nailing it down.
I am reminded that Niels Knudsen is THE expert on Phase/C1 colour issues, with specialised knowledge of how CMOS sensors perform. So it is doubtful that I can add anything significant to what is already known. But maybe, just maybe.....
I am wondering if Niels' finalised Phase icc camera profiles may be biased in some way, perhaps optimised for studio-like conditions - in particular with low ISO settings. Much of my photography is done in relatively gloomy conditions at 3200 ISO, give or take. So, it occurs to me that a camera sensor may have a different icc profile at high ISO, such that the yellow-orange zone may be overly-saturated compared with the blue end of the spectrum.
Has anyone any information on this? Perhaps aficionados of Raw Digger? Or users of QPCard?
Thank you.
Peter.0 -
Hi Chris
Your A7r image (well-balanced colours by the way) - are you saying C1 has a special icc profile for A7r, and did you use the special C1/Sony software?
If so, wouldn't it pay Phase to start doing some v2 profiles for Cannon and Nikon? And how does a v2 profile differ from a "Generic" profile?
Regards, Peter.0 -
[quote="SFA" wrote:
[quote="RichardT" wrote:
All using default settings and exported to JPEG.
Richard,
Could you run a comparison for the different Film Curves for the 7D2 profile please. (I assume you you have left the default set to Auto for the C1 examples and that Auto has elected to work with Film Standard?)
Grant
7D2 + Extra Shadow
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1104_CO_7D2_extra_shadow_zpsworprulm.jpg
7D2 + Film Extra Shadow
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1104_CO_7D2_film_extra_shadow_zpsptakrzkp.jpg
7D2 + Film High Contrast
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1104_CO_7D2_film_high_contrast_zpsixpaaosq.jpg
7D2 + Film Standard
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1104_CO_7D2_film_standard_zpsxttix5eu.jpg
7D2 + High Contrast
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1104_CO_7D2_high_contrast_zpsodhjs6ya.jpg
7D2 + Linear Response
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1104_CO_7D2_linear_response_zpsqpyoxggs.jpg0 -
Another example (these are from the image I submitted with my support case)
DPP
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1094-DPP_zps9ev7btwt.jpg
CO + 5D2
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1094_CO_5D2_zpsdiwts7w6.jpg
CO + 7D2
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1094_CO_7D2_zpsvcqvamv2.jpg0 -
And to show that it's not just buildings in Oxford that are affected...
DPP
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_0865_DPP_zpsbvytx3n6.jpg
CO + 7D2 profile giving the sea an unhealthy orange tinge
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_0865_CO_7D2_zpsbjueskfs.jpg
It's easier to compare the images if you click on one and then step through them on photobucket.0 -
Hi Richard
Thank you for introducing me to Photobucket! Unfortunately my upload speed is somewhat sluggish but I may give it a try.
I have analysed your jpg images created by you using default software settings. The average RGB readings are as follows:
______________________DPP_________________CO8-7D2____________________ACR
All________________173 167 158____________164 155 146_______________160 154 148
Sky only___________244 247 251____________240 242 245_______________239 240 243
Building only_________98 82 61_______________85 62 41__________________76 64 47
Column only________152 129 96_____________144 105 68________________125 107 81
These readings are, of course "absolute", independent of screen settings. (Sorry about the layout - I can't find a means for sending a Word doc attachment.)
In my work I have settled on DPP as giving me a reliable and consistent starting point. In my experience Silkypix 6 gives an equally good rendition (Canon 5D2). And until Phase gets C1-8 sorted out for Canon I have reverted to v7.1.3.
So, with DPP as our reference point perhaps the most significant point arising from these figures is that what looks like an overly red building in CO8 is not due to an overactive red channel but underactive green and blue channels.
The ACR data seems to have its own "take" but it is closer to CO8 than DPP.
As a matter of curiosity I tried using the CC color balance tool to make the CO8 image look more like DPP. It is impossible!
Food for thought - and time for Phase to start reporting back to us.
Regards, Peter.0 -
Richard,
Thanks for posting the comparisons. Most interesting. Especially the default DPP which seems to come across quite a bit lighter and with less contrast than any of the others. Is that what you see on screen too prior to jpg creation?
Grant0 -
Yeah, that's the exact same browny-orange colour shift I'm seeing in my 7D Mk II conversions.
I shot a game of football (soccer, if we must) recently - gorgeous Winter sunlight, which Photo Ninja, Lightroom and Optics Pro all rendered beautifully: the same files in Capture One 8 were so "off" that I honestly thought I was looking at them in the wrong colour space - they were so wrong-looking tonally (particularly in the flesh tones) compared to conversions from the other converters that I really believed I'd exported them (and was viewing them in Capture One) in Adobe or ProPhoto RGB rather than sRGB.
(No need for anyone to tell me where I'm going wrong with my white balance or colour management - I know exactly what I'm doing here, and I'm fully colour managed: I'm just using this reference as a familiar analogy to emphasise my point).
I'd be very happy to get results akin to the Sony A7r example above. interesting, is it not, that - Phase One having a commercial relationship with Sony - Capture One provides excellent profiles for Sony cameras?
Just maybe Phase One is getting its Sony profiles from Sony (I'm going to go ahead and assume that's the case - or at least that Sony is "contributing" in some meaningful way to the generation of these profiles. Crazy not to, I'd say).
If I'm right, it further demonstrates that the problem we're discussing here comes from Phase One's profiling regime, a problem from which Sony users are isolated by dint of the probable source of their profiles...0 -
Hi Peter,
thanks for your analysis - the figures make interesting reading. I agree, DPP is currently giving a much better starting point than CO8.
Hi Grant,
I can confirm that the image I see on screen before jpeg creation is very similar to the contents of the jpeg.
Hi Keith,
I agree it would be good to get results for the 7D2 that were as good as the Sony example.
Hope we have persuaded Phase One that the handling of other cameras in v8 is not acceptable.
Regards,
Richard0 -
[quote="Peter" wrote:
Hi Chris
Your A7r image (well-balanced colours by the way) - are you saying C1 has a special icc profile for A7r, and did you use the special C1/Sony software?
If so, wouldn't it pay Phase to start doing some v2 profiles for Cannon and Nikon? And how does a v2 profile differ from a "Generic" profile?
Regards, Peter.
Hello Peter,
The Sony A7r "standard" profile could be a generic v2 Phase one profile (there is a generic v1 profile, but that is nowhere as good and above all: reliable as the second "standard" profile).
As to whether Sony itself has had any contribution in the high quality of this profile, I don't know. But for some reason it just doesn't follow the (recent) trend of Phase one to favor a certain warm, reddish look, and in the meantime succeeds at what the Phase one vintage look fails at: give reliable, consistent ánd beautiful colors across a broad range of images.
I must say b.t.w., that the samples of the Canon 7DII look particularly as if they suffer from a cast. Perhaps collect some samples where crucial colors shift, and enter them in a support case. I am sure a consequent v2 profile will be much better, usually the v2 profiles really are.
Chris0 -
[quote="ChrisM" wrote:
As to whether Sony itself has had any contribution in the high quality of this profile, I don't know.
Oh, I don't know either, but I'm a great one for choosing the explanation which best fits the available evidence - and this does just that.
In any event, surely - as Richard suggests - we've made the case by now. And isn't it interesting that John up the page - a Nikon user - has joined in to express similar concerns?
Further evidence that a camera-by-camera, support-case based solution is the wrong way for Phase One to go about addressing this issue.
John, one thing I will say: we'll never get "accurate" colours out of the box with Capture One - that's not what the software is about, and Phase One is on record as saying that they aspire to "pleasing", not accurate, colours.
Fine, most of us would be happy enough with that (back in the day this was arguably Capture One's key selling point - its uniquely characteristic, and extremely appealing, colour rendering) - but as of right now, "pleasing" is precisely what we're not getting.0 -
Peter,
Thanks for taking the time to generate the individual colour analysis figures from the various samples, very interesting reading !
Looking at the various sample shots from a 7D2, they're identical to the sort of problems I've had with colour from Nikon's and its bloody awful. I think Keith mentioned when he looked at some files they were so flat he wondered if the colour profile was correct which is exactly what I thought too when I was looking at mine at one point!
Keith - I take your point about colour and Capture One but as you say we're all saying the same thing here and the issue is not camera specific, its a problem with the software !
I'd hope Phase One acknowledge acknowledge this issue and reassure their customers they're pro-actively doing something about it.
John0 -
The postman arrived early today. A letter from Phase? Yes!
Dear Peter
We are grateful to you and fellow users of Cap1 for drawing our attention to some colour problems when processing Canon, Nikon and Pentax raw images, and for the various ideas you have put to us.
We have rechecked the colour settings of our laboratory equipment and found nothing amiss. We are now retesting the 8.1 software using samples submitted in the "Part 2" thread, as well as our own test images.
Our initial findings are that there does seem to be a colour rendition problem and we are working hard to rectify this. We hope to release an improved version of Cap1.8 by the end of the month. In the meantime we recommend that you revert to v7.??? which is compatible with all but the very latest cameras.
As a gesture of appreciation for all the work you and your colleagues have done to support this thread we are pleased to allow each of you a free upgrade to Cap1 v9 when it is available.
We apologise for this uncharacteristic failure to live up to the high standards that we at Phase always try to maintain.
--------------------------------------------------------------------------------------------------------------------------------------------------
I opened my eyes just in time to see a pig flying away.
Well, we can all dream.
Peter.0 -
Not sure if it will help, but here is a 7D2 photo of x-rite colorchecker converted to jpeg using DPP and CO8 using default settings (Phase One support wanted one of these for my support case).
DPP
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1213_DPP_zpsjrbvcefu.jpg
CO8
http://i1381.photobucket.com/albums/ah220/Python54/phase_one/IMJ_1213_CO_7D2_zpst39hp4st.jpg
Regards,
Richard0 -
[quote="Phase One" wrote:
We have rechecked the colour settings of our laboratory equipment and found nothing amiss.
I suppose they're bound to say this - but (like many of us, I imagine), I've got an IT background, and I know fine well that something can be "to spec" and still not work properly if the spec itself is shonky.Our initial findings are that there does seem to be a colour rendition problem and we are working hard to rectify this.
So we were right all along. Kinda suggests that the precipitous slamming shut of the earlier thread on the subject was - shall we say? - ill judged (and which - I don't deny - ticked me off royally).As a gesture of appreciation for all the work you and your colleagues have done to support this thread we are pleased to allow each of you a free upgrade to Cap1 v9 when it is available.
I wonder if that really means everyone who has actively pushed on this topic? It's a welcome gesture though, if Phase One is good to its word here and doesn't try to exclude some of the contributors on the basis of some arbitrary line in the sand.[quote="Peter" wrote:
I opened my eyes just in time to see a pig flying away.
Let's give 'em the benefit of the doubt, Peter - at least they're not peremptorily dismissing our concerns any more.
And well done to everyone who has backed this effort - it would have been easy to back down under the pressure of Phase One's earlier continued insistences that there was no problem, and its apparent unwillingness even to entertain discussion on the matter.
A good result - maybe.0 -
[quote="Keith Reeder" wrote:
And well done to everyone who has backed this effort - it would have been easy to back down under the pressure of Phase One's earlier continued insistences that there was no problem...
A good result.
Keith,
I rather suspect you might want to re-read Peter's post. Especially the end.
Grant0 -
I did, Grant - I read it as Peter not holding out much faith that Phase One will, after all, do right by us, not that they haven't been in touch.
This much I do know: if Phase One doesn't step up and do right by us here, I'm done with Capture One.0 -
Hi Richard
Your x-rite images. I have scanned across the top 8 patches, from red through green to magenta.
The data (average RGB readings) are as-listed. First column is DPP, second column is CO8-7D2:
255 106 117___244 100 100 red
254 177 58____255 177 56 yellow1
254 226 56____245 219 80 yellow2
70 232 170_____10 198 148 green
111 216 248____71 202 233 blue1
107 182 255_____6 173 244 blue2
191 135 216___173 137 210 purple
251 95 171____248 66 149 magenta
There are a few glaring mismatches here. If Phase cannot or will not acknowledge that they have a problem with 7D2 rendition then I despair!
Peter.0 -
Interesting.
Comparing through Firefox on my notebook (2 tabs both with the post open, align vertically and then click between the tabs for an overlay) the processing is quite different DPP to CO8.
CO8 seems to have produced more contrast, more saturation (notable in the blues and "greens") and is apparently sharper (for whatever reason(s)
On the colour chart the red/pink colours seem quite consistent with small variance and the yellow maybe changes just a touch. But the blue and green squares and thise influenced by them generally have rather obvious changes which are also evident in the cloth (microfibre?) under the checker card, the foliage outside the window (though that could be a yellow related change in some way) and the grey areas and shadow on the right of the image. (DPP is grey, CO8 is blue grey)
However this is hardly a conclusive analysis even by "eyeball" standards.
Were you using the "Film Standard" curve?
What do others see?
Grant0 -
Hi Grant,
Another way to see the differences is to click through one of the images to photobucket and there use the arrow keys to step backwards and forwards between the two images. To me this confirm Peter's figures, especially the green in the top row!
Regards,
Richard0 -
[quote="RichardT" wrote:
Hi Grant,
Another way to see the differences is to click through one of the images to photobucket and there use the arrow keys to step backwards and forwards between the two images. To me this confirm Peter's figures, especially the green in the top row!
Regards,
Richard
Hi Richard,
I did that with your other sample files.
I've used Photobucket for years so I have a long standing doubt about some of its rendition capability though no doubt it is much improved these days. What I don't see on the bucket (maybe I would if I signed in) are the file names for your original samples which made it slightly frustrating doing the comparisons. Less of an issue with only 2 images of course.
Aligning in a browser offers the same sort of comparison but retains the ability to see your in-post annotation and the option to align different variants for the comparative switch. Not so easy to do in bucket without some messing around - maybe not at all possible?
Have you also got some lens correction differences going on in the jpgs?
Grant0 -
[quote="SFA" wrote:
I've used Photobucket for years so I have a long standing doubt about some of its rendition capability though no doubt it is much improved these days. What I don't see on the bucket (maybe I would if I signed in) are the file names for your original samples which made it slightly frustrating doing the comparisons. Less of an issue with only 2 images of course.
Yes, it's annoying they don't show the filenames on the page; however they do appear embedded in the URL in my browser's (Pale Moon) address bar.[quote="SFA" wrote:
Have you also got some lens correction differences going on in the jpgs?
Grant
Not intentionally - the lens settings are the defaults for each application. CO8 shows the correct lens (EF-S 17-55) but DPP said "Lens Data: No". I have since followed the option to download the specific lens data and it now says "Lens Data: Yes". However, when I re-generated the jpeg, the geometry of the image was unchanged.
Sorry, I forgot to answer your earlier question - Capture One curve was set to "auto". Comparing the auto setting against the actual curves it does correspond to "Film Standard" in this case.
Regards,
Richard0 -
Thanks for the info Richard.
True about the filenames in the browser URL. I think I saw the section before the file extension and looked no further left! Doh...
Interesting that the Canon DPP lens data load makes no difference.
It might also be interesting to use DPP and CO to output the Linear he file to avoid any additional interpretation issues after de-mosaicing.
Also to know if the CO lens adjustments include anything for generic colour cast adjustments as one might have in an LCC file.
So many components in the process that just might in small ways tweak things more as a sum of adjustments than they would seem likely to do individually.
All rather interesting.
Thanks for your further efforts.
Grant0 -
[quote="SFA" wrote:
Interesting that the Canon DPP lens data load makes no difference.
I had another look at DPP and found that I should have scrolled the tool panel down a bit and ticked the "Distortion" checkbox. This appears to apply the lens data and the resulting jpeg then looks more like the one from CO8 (probably about half between that and the DPP jpeg I uploaded).
I am afraid any further investigations will have to wait as I have other things planned, including actually making use of my camera gear taking photos! 😊
Regards,
Richard0 -
[quote="RichardT" wrote:
I am afraid any further investigations will have to wait as I have other things planned, including actually making use of my camera gear taking photos! 😊
Regards,
Richard
Good grief man. You are not supposed to keep adding to the problem workload you know. You haven't really understood this post processing lark at all have you?
😂
Enjoy your day.
Grant0 -
I'm glad to see Phase One have acknowledged they have an issue and the results from the color checker passport will reinforce the position. Hopefully they'll keep us informed as to whats happening as we shouldn't have to rely on a letter to an individual being posted on this forum to tell those of us who've spoken out on what is a very significant issue in their software.
I've moved to Nikon's NX-D for RAW processing which is Silkypix with the all important Nikon algorithms added. I shan't touch Capture One again until its clear they've sorted out these issues and if they choose not to communicate what they're doing they'll loose a customer !!
John0 -
Hi John
Please can you tell me where Phase has acknowledged it has a problem? Perhaps in a support case?
Thank you, Peter.0 -
Peter,
I was referring to your previous post when you advised you had received a letter from Phase ??
John0 -
Hi John
I am so sorry. The "letter" is a fiction, my way of tempting Phase to raise their head above their wall of dignified silence.
We mere Phase customers have had no feedback at all. So I tried to show what we could reasonably expect from a supplier whose product appears to have a serious flaw. This will probably backfire (on me) because I don't think Phase will appreciate this kind of humour.
But it has succeeded in keeping the show on the road, with new information to analyse.
One lesson in all this, one definite lesson, is that if you like Cap1 in principle (and I do) you should consider switching from Canikon to Sony!
Peter0
投稿コメントは受け付けていません。
コメント
225件のコメント