I finally got myself a colour chart and can now use this tool.
However, I took two shots of the card today: one works exactly as advertised and the other does not. I could not find any information on what is wrong with the failing image such that it just says this:
I have tried:
Removing ALL adjustments.
Removing all except distortion adjustment (it’s pretty obvious, so might make a “non-square” target that is hard to find).
Increased only exposure to brighten the target.
Noting a slight shadow across two of the swatches, I carefully used local adjustments to all but eliminate the shadow.
Re-framed the target to shrink the width to avoid the worst of the shadows.
None of this made any difference.
I will need to be more careful with casting shadows — apparently my target (DataColor Spyder Checker 24) is really not intended for hand holding, as its margins are very slim.
This is the shot that worked (it even has default adjustments applied).
I am at a loss other than to assume the slightly deeper shadow (cast by my hand) is enough to throw it off. I was, at least, able to use one of the grey swatches to set white balance.
I got a colour chart a few months ago and found that the lighting does need to be bright enough and even.
It was very frustrating at first “It’s lined up… why isn’t it working?”. In my case it’s because I tend to underexpose (old bad habit I started to change because of this).
For my use I found it best to hang the chart from something. But being paranoid about damaging the colours, I got one in a flip-case with string (on sale in November).
If I were to get new charts for some reason, I’d just get the charts and make a sandwich-flip-case out of thin ply-wood.
Calibration behavior sometimes fails for reasons that are not easy to understand. But I always managed to get a profile after a few tries.
In your case, it is probably indeed the shadow of the hand on one of the boxes. I also noticed that the order of positioning of the points can have an impact depending on the orientation of the chart in horizontal or vertical.
Generally the charts have crosses at the corners to position the grid. This is not the case for this Datacolor charter. On other charts, these crosses are closer to the colored squares than the edges of the chart, and the edges are wider. As a result, there is less risk of the hand shadow causing a problem
I suggest you start your calibration again by positioning the points near the exterior angles of the color patches: this will offset the measurement squares (white frame) in relation to the color patches, but there is no impact as long as that the white frame remains in the patch. This might help resolve the problem.
Yes, but this seems linked to the sensor’s high ISO capacity.
For example with a Sony A7RIII, I create a profile up to ISO 32000 without any problem. But it is no longer possible beyond that.
In this case 32000 iso is the last normal iso value, then up to 102000 iso, these are “extended” iso values.
(PL7 Windows)
