Can't perform color calibration due to clipping

I only took one shot with the color card and now it seems I cannot use it. It appears that elsewhere (outside the card) in the image I have some clipped values, but within the card itself everything is within range. Yet I get the message “Chart position error and/or clipped values detected in RAW file”.

I tried to crop out just the color card and export a full resolution TIFF of that, but PL does not seem to be able to create a calibration from a TIFF. Not sure how to go about actually cropping (destructively, so to speak) a RAW…

This seems like a flaw in the application logic. It should only be concerned with clipped values inside the confines of the calibration card part of the image shouldn’t it?

Any workarounds for this?

just take a another shot…

as a general note - unless spectrum of your illumination is really far far away from any flavor of daylight from dusk to dawn or from any incandescent light or from xenon flash impulse ( think odd spectrum like for sodium vapor lights for example

) there is ABSOLUTELY NO point to use substandard profile generation ( which produces a simple single illuminant pure matrix DCP profile ) in DxO PL7 …

PS: if you really , really need to do this, you can post your raw file and somebody can fix the raw data (and/or relevant exif tags) inside to bring the DNs in clipped areas within acceptable range… there are number of ways to do this… if indeed the matter is with some specular highlights being clipped outside of target patches and not something else whatsoever

That’s not really practical advice. The scene and its light are long gone at this point. In other cases with humans the moment is gone, etc. Reshooting and recreating the exact same conditions is almost never going to be viable.

I don’t know enough about the technical details to debate this on that front, but I can say that when I set my WB to flash and I use a Nikon speedlight with a Nikon camera the colors are still not as accurate and vivid as when I add in the cal from the card. I shot a scene the other day with lots of pinks and yellows and before the cal it looked like it was artificially lit (which it was - all light from the flash) and had a bit of a blue cast. But after the cal everything was perfect. So I definitely plan to keep using this whenever practical.

To my original point though, there is no reason why the calibration process should be rendered inoperative due to other parts of the image having values out of range. That is something that future versions of PL should be able to address in order to prevent people from being “stuck” days later in post when they realize their cal isn’t useful (without extraordinary measures at least such as RAW hacking).

Looks like an oversight to me, it should be reported as a bug.

As a workaround, you can use other free software to create a dcp profile, either the free Adobe dng editor, or calibrite also offers a free software. You can then import that created dcp profile and use it in Photolab. That’s what I’ve been doing in previous versions of Photolab.

1 Like
  1. unless you own a proper spectrophotometer and measure what you shot you shall not use the word “accurate”

  2. matrix profiles in principle can’t be accurate (as in reproduction) unless camera’s CFA satisfy certain conditions and consumer cameras don’t have that

  3. “when I set my WB to flash” - use WB measured off ~uniform across 380…400nm to 700…730nm spectrum reflectance target like white thread sealant teflon tape instead of assuming that WB in DxO’s drop down list is precise - plus DxO has a number of profiles - default might not be designed for any accuracy at all

  4. “So I definitely plan to keep using this whenever practical.” - if you want a general purpose profile with any claim to accuracy you need to try to create lut based profile using other tools ( and remember that profiles created with targets where just one patch barely is outside of sRGB gamut - cyan patch that is - can not be accurate for truly “vivid” colors )

1 Like

there are much better tools to create DCP profiles - suggestion was in case you do not want to learn them … again if you want quick fix (if the matter is just to nip the clipping outside of patches) from community post the raw file, if you can live w/o quick fix - file a bug report with DxO and wait for a fix from DxO ( good luck w/ that )

I don’t want to spread misinformation so I’ll add more to this… I had the idea to lower the exposure to see if that changed anything. Even with the default +0 exposure I couldn’t find any R G or B values beyond 226 in the color chart area so that didn’t seem like it should be a problem. Still, I dropped the exposure to the point where the highest values were in the 180s and that didn’t change anything either. I started moving the corners around a bit and found one spot where the “Save and apply” button actually lights up blue and allows me to save a calibration.

So now I’m not sure what it going on here. There are still some clipped pixels showing elsewhere in the image even at this lowered exposure, so that doesn’t seem to be a total disqualifier. Yet the pixels in the calibration area are way, way below clipped levels and it’s still hit or miss (much more miss than hit) as to whether it can create a cal.

So do I perhaps have a stuck pixel within this part of the image and PL is getting tripped up on that even though I’m unable to see it on screen? Not sure…

In any case, all I can say for sure is it seems to be worth taking the time to drag the corners around and trying different overlay positions. Even if your squares are all inside of the cal squares there may be different results based on slight position changes.

See this post:
(same chart, same problem ? )

1 Like

Thanks for that link, I’ll give it a read/try!

Thanks for reminding me why I was considering leaving the forum. Decision made, ye of no public profile.

One shall not use the adjective proper if one wishes to be accurate.

As to the profile creation (with acceptable precision), using the Spyder Photo I can confirm that aside from even and bright enough lighting, the corner points also have to be lined up a certain way. The small squares always have to be inside the colour squares but not necessarily aligned evenly.

@OzarkNerd you don’t need to have the watch in the shot for creating the profile. Do your shot with the chart, then replace the chart with the watch. Take your shot of the watch and use the profile from the photo with only the chart to apply to the shot of the watch.


how sensitive we are down under …