EDIT: Which result is not correct? Looking at the scans, I thought that the Kodak paper was wrong, however your post’s title says that the Epson paper is the problem… I presume that you have press sheet results that “match” one of the proofs too.
Just curious, when you changed the paper, did you re-align the inline SpectroProofer to the new media using the front panel of the Epson? You would then probably have to power off the proofer and then power it back on again due to a bug in KPS that does not like the ILS being realigned without a power off/on cycle.
I don’t know why you are getting such different results. Looking at your simulation devicelink and proofer profile, all is OK. As these are certified prints, the system will not let you certify unless the calibration is less than 7 days old. What is the max and average dE values for the calibration? As the colour bar requires that you use the correct signed simulation and proofer profiles, if a profile was incorrect then you would not be able proof as it would kick the job out.
Have you done a nozzle check? It could in theory be remotely possible for the patches to pass certification, while the job itself could suffer with a colour issue – however that would be a very strange case.
Your 7m22y background in F39 = L*91 a*2 b*14 (AbsCol) or L*95 a*2 b*16 (RelCol), which are both warm hues. The job proofed with the Epson profiles/media is warm. The job proofed with the Kodak profiles/media is cold (presuming that both images were captured in the same conditions). About 18 months ago Kodak sold off their Matchprint media to Ilford who now manufacuture it as Kodak are no longer making media, so perhaps this is a media issue or you need updated media configuration file, simulation devicelink and proofer profile. If the new media was that different, then the certification should fail, so I doubt that is the case.
There are many versions of the Kodak Media Installers and many versions of the colour bars, proofer device profile and simulation devicelink profiles. Perhaps you should try an earlier or later set of assets. Looking at what I currently have installed in Proofer Administrator for our 9900:
T1SM240P 7x14 F39U wedge v3 ILS U v8.zip (colour bar/media wedge)
Mx9_T1240_7x14_F39_3v5_m_U.dvl (simulation devicelink profile, signed and linked to the media wedge)
Mx9_T1SM240P_7x14_3v3_U.icc (proofer/media profile, signed and linked to the media wedge)
I guess the next step would be to see what proofer device values are being generated and then work out what L*a*b* values they equate to. You would need to RIP both jobs again and hold them in the queue, perhaps using a scatter layout – they don’t need to be printed. You would then open up a high resolution sheet preview and use the eyedropper to find the CMYK proofer colour values being generated by the Epson paper and Kodak paper profiles.
Stephen Marsh