| Author | Message | ||
     Charles McMillan New member Username: Cfmcmillan Post Number: 5 Registered: 10-2004 |
When I do a match reference on the MacBeth target in the Chrome 2000 D65 working space I get very reasonable results. However, when I am working in the Chrome 2000 D50 space, the results are WAY too blue. Any idea what's going on? Charlie | ||
| Dieter Mayr New member Username: Dieter_mayr Post Number: 82 Registered: 11-2004 |
Charles Did you convert the images into the working colorspace ? Maybe you can describe your workflow, I was not able to reproduce your results. Dieter | ||
| Charles McMillan New member Username: Cfmcmillan Post Number: 6 Registered: 10-2004 |
Dieter - My work flow was a follows: Path 1 - Workspace in PW Pro set to Chrome D65 asumed input file in sRBG space Load a jpeg of the MacBeth Color Checker in sRGB space Use Match Reference - the resulting photo looks good Path 2 Workspace in PW Pro set to Chrome D50 asumed input file in sRBG space Load the same jpeg of the MacBeth Color Checker in sRGB space Use Match Reference - the resulting photo looks awful. I'd be happy to send the input photo if that would help sort things out. Charlie Charlie | ||
| Dieter Mayr New member Username: Dieter_mayr Post Number: 88 Registered: 11-2004 |
Charles I did some experimentation now. When i use a sRBG image of colorchecker and convert it into woking colorspace, i get good results, no matter if D65 or D50. The problem occurs when i did NOT convert, and put the image in sRGB space into Match Reference Transformation. Then i get bad results. But that dont matter if D65 or D50, too. The input image was sRGB, but the output was in the working colorspace, and i think this causes the problems (but i cant explain why). If you want, send me your picture, i will convert it and send you the results. You find my e-mail adress in my profile. Dieter | ||
| Charles McMillan New member Username: Cfmcmillan Post Number: 7 Registered: 10-2004 |
Dieter - I sent an image via email, but the email program may have had trouble sending it. If you don't get it, let me know. Thanks, Charlie Charlie | ||
| Dieter Mayr New member Username: Dieter_mayr Post Number: 89 Registered: 11-2004 |
I experimented with Charles images and had a mail discussion with him abt the results. With his permission, i copy it here for further discussion The pictures reffered to you can fing here : http://photos.yahoo.com/dietersbg68 in album public. All files are with theyr profiles. My mails i sent with the pictures: Charles Here are my results of my experimentation TestCharles is your original file TestCharlesColorBalance I correted with Color/Balance removing casts with custom settings for highlights, midtones and shadows (eyedropper on the white, black and a midtone field of colorchecker) TestCharlesConvD?? are converted when opening from sRGB into working colorspace (Chrome2000 D65 / D50) TestCharlesNoConvD?? same as above, but no conversion when opening. The D50 images seem a bit yellow, but that seems to be normal to me, because my monitorprofile is set to D65. I dont think any of the images looks awful, as you mentioned. Let me know please what you think of my results. Charles reply: Dieter - Your results look the same as mine - I have my monitor calibrated for D50. However, the colors are very close to right when using the D65 working space and are clearly too yellow (awful in my parlance) when using the D50 space. Simply doing a white balance correction, as you did, does much better than using the full correction in D50 space. Since the purpose of using the color checker is to get very accurate colors, particularly for portraits - skin tones and the like- too yellow makes people look jaundiced. What I can't figure out is why the color space would matter. For these two color spaces, the gamut is the same - it's just the white point that's different. It makes me suspicious that there's a bug burried in the program somewhere. The general rule I've used in both experiments and programming is "the things you don't take the time to understand will reach around and bite you." I'm concerned that we may have an indicator of a problem here. Regards, Charlie My reply: What i was thinking of, i dont have much expierience with colorchecker, can reflected light from the wooden surfaces be a problem ? I mean if you have natural or artificial light on the furniture, but reflected light on the colorchecker, i think this could cause a colorcast, because the full corrected Images are much more yellow then the only whitepoint corrected. But it doesn't explain the difference in colorspaces, of course. If its ok for you, i will copy our conversation here to PW-forum, and also the pictures i sent you to be available so others can tell us their opinion, too. Please tell me if thats ok with you. Have a nice weekend Dieter and finally Charles answer: Dieter - It's certainly fine to put these up for the discusion group - that's where I started it -I'd suggest putting it with the thread that's already started on this so it all stays together. Light from the reflected surfaces could be a problem. It's easy to get rid of that though - just put a white card down below the color checker. Another alternative is to put down a black cloth. However, that should only be an issue for getting the color right in an absolute sense, it should make a difference which color space we are working in. Regards, Charlie |
