Matching prints from Epson P900 and 7570

[Doug Gray]

Of course, it is easy to cripple an image for one printer that wouldn't be crippled on the other.

Nonsense. The large majority of images I've seen that aren't saturation pushed, especially in ProPhoto, are within gamut of most printers. Well, except for laser printers. I do repro work and it's unusual to repro an image that can't be printed on a wide variety of printers. Biggest issues are paper white point and printer black point which can be a go or no go. And it's pretty critical that the prints match the original.

[digitaldog]

Nonsense. The large majority of images I've seen that aren't saturation pushed, especially in ProPhoto, are within gamut of most printers. Well, except for laser printers. I do repro work and it's unusual to repro an image that can't be printed on a wide variety of printers. Biggest issues are paper white point and printer black point which can be a go or no go. And it's pretty critical that the prints match the original.

The large “majority” is specifically how many?
And all those you've seen, you've plotted the 3D gamut, really? 🤥
And I'm posting nonsense; yup.

[Doug Gray]

The large “majority” is specifically how many?
And all those you've seen, you've plotted the 3D gamut, really? 🤥
And I'm posting nonsense; yup.

I much prefer Matlab to examine distributions where an image is OOG. Photoshop is crude and won't show OOG for printer type ICC profiles until the OOG reaches about dE6. 3D gamuts can be useful but it's hard to read the specifics and trivial in Matlab. For a quick view, flipping between converted images as I outlined earlier works well to spot areas of gamut issues. Keeping in mind that if your working space is larger than your monitor, you need to desaturate in settings. Especially for ProPhoto. While unpleasant to look at, desaturated images are still useful for identifying any areas of issue.

As for the idea that my conversion wrecks images. It only clips color in that part of an image that can't be printed on either printer. If it can't be printed on both then you can't print the image on the two printers and get the same print.

[digitaldog]

I much prefer Matlab to examine distributions where an image is OOG. Photoshop is crude and won't show OOG for printer type ICC profiles until the OOG reaches about dE6. 3D gamuts can be useful but it's hard to read the specifics and trivial in Matlab. For a quick view, flipping between converted images as I outlined earlier works well to spot areas of gamut issues. Keeping in mind that if your working space is larger than your monitor, you need to desaturate in settings. Especially for ProPhoto. While unpleasant to look at, desaturated images are still useful for identifying any areas of issue.

As for the idea that my conversion wrecks images. It only clips color in that part of an image that can't be printed on either printer. If it can't be printed on both then you can't print the image on the two printers and get the same print.

That's all just lovely but doesn't answer my simple questions:

The large “majority” is specifically how many?
And all those you've seen, you've plotted the 3D gamut, really? 🤥
And I'm posting nonsense; yup.

[Doug Gray]

Mostly my repro work is photos or documents. For photos I use an Epson V850. It has severe issues wide area crosstalk. I wrote a program which removes it. There's a good description of the problem at github and why ICC profiles can't correct it. For larger repro's, I photograph the artwork. This involves quite a bit of code as I need to adjust for the illuminant with a reference white sheet. Then dimension adjust and finally correct for color shifts by spot checks with a spectro. Messy. Mostly I work in Abs. Col. and the only problems involve black point issues because my large format printer, Epson 9800, has pretty high black points. So that limits things if I have to have an image that can be matched printed on the Cannon Pro1000 and 9800.

https://github.com/doug3236/scanner_refl_fix

You might also be interested in this utility that aggregates RGB patches for profiling. This allows easy comparison of ICC accuracy by aggregating multiple patch sets, randomizing placement, and printing them all at once.

https://github.com/doug3236/i1Patches

There's also a utility for making B&W ICC that show softproofing correctly using tinted advanced B&W print modes.
https://github.com/doug3236/ABWProfilePatches
 

[digitaldog]

I see you are not going to address my simple questions.
Adios then.

[Doug Gray]

I see you are not going to address my simple questions.
Adios then.

Which part of why I don't bother with 3D gamut plots did you not understand? I don't waste my time with 3D plots.

As for how many pictures out of the 100-200 I've repro'ed, I don't recall any that were materially OOG outside of BP and there weren't many of those (< 10%) . But there may have been a couple on the edge with color saturation. I just don't recall them. However, these were mostly printed photos. By definition, since the originals were printed years earlier, their gamut was limited by the media.

If you want to make an argument, provide an image (an actual scene w/o synthetic colors) that shows degradation when tripped through the two conversion processes I outlined. Defend your position. You said this:

Of course, it is easy to cripple an image for one printer that wouldn't be crippled on the other.

Really? I'm open to examples.

[digitaldog]

Which part of why I don't bother with 3D gamut plots did you not understand?

The large “majority” is specifically how many?

100-200 you've repro'ed? There is hardly a significant sampling with zero data on capture, subject, processing, and proper analysis. When you've spent 5 decades + making images, 3 of them digitally, and you have many tens of thousands under your belt, let's talk about a large majority of images you've seen that aren't saturation pushed, especially in ProPhoto, that are within the gamut of most (not all?) printers.

[digitaldog]

Really? I'm open to examples.

There's a very old video I did you can dig up and view. It's all there.
Tonight is the Super Blue Moon (as well as the DNC). I don't know about you, but I'm a photographer. I'm going out to make images. Give it a try.

[Doug Gray]

There's a very old video I did you can dig up and view. It's all there.
Tonight is the Super Blue Moon (as well as the DNC). I don't know about you, but I'm a photographer. I'm going out to make images. Give it a try.

Well, your stuff is quite good. I consider it the best on the web in terms of general photography and color printing. I have the Canon EOS-1Ds II, and R5. Shoot raw. Gave away my old EOS-1Ds 5 years ago. Mostly I enjoy rural landscape photos.

That said I'm a retired techie/coder professionally. I got deeply into the construction of ICC profiles when looking at some of the differences between i1isis profiles and the OEM ones. Then I got irritated at the design of the V850 scanner. The problem of light that is reflected from surrounding regions causing erroneous readings of Photo IT8 charts for making profiles bugged me. It might even be intentional because it also has the effect of increasing local contrast. Great technique in Photoshop for making an image "pop" more but it's not a realistic scan and it would have been easy enough for the manufacturer to fix. But I found a way to fix it and a side effect is that profiles from IT8 charts produce more accurate results based on spectro spot checking the IT8 charts.