Revision 11826 disables the use of high level images for images where we downscale. Regression testing of this change shows MANY differences. Some are small (an overall lightening of the image), but others are much more significant (rendering now happens without halftoning). It appears that the new rendering (i.e. that without high level images) is correct, which means that there must be a bug in the high level image code. I will attach a patch to revert 11826 in a moment that can be clusterpushed to see the differences.
Created attachment 6810 [details] patch.txt Simple patch to demonstrate the problem (reverts revision 11826, and thus causes high level images to be used for downscales). See in particular: tests_private/comparefiles/Bug688728.pdf.pgmraw.300.0 tests_private/comparefiles/Bug688728.pdf.pgmraw.300.1 tests_private/comparefiles/Bug689520.pdf.pgmraw.72.0 tests_private/comparefiles/Bug690321.pdf.pgmraw.300.0 tests_private/comparefiles/Bug690321.pdf.pgmraw.300.1
This is really a clist issue, but I'll consult with Michael w.r.t. the colorspace confusion. Note that this is a regression in 9.00 from 8.71. Wonder how this snuck through. Making this a P2 since it will affect customers that use 'gray' devices, but hopefully there aren't that many.
(In reply to comment #2) > This is really a clist issue, but I'll consult with Michael w.r.t. the > colorspace confusion. > > Note that this is a regression in 9.00 from 8.71. Wonder how this snuck > through. > > Making this a P2 since it will affect customers that use 'gray' devices, > but hopefully there aren't that many. Sorry, I suggested Robin assign it to Michael because during playback the image is being mapped through the pdf14 device which in turn is not setting up the pdf14gray device properly. It is setting the number of component correctly but max_gray was incorrect. You can see this right away stepping into the first begin_image device call from the playback. Just FYI.
As per my email to tech "banding/page mode bugs", closing this file as I cannot reproduce a problem.