Pages: [1]   Go Down

Author Topic: Photomechanic + FastPicViewer seem to act weirdly with NEC PA profiles  (Read 1973 times)

WombatHorror

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 299

EDIT: actually i'm noticing a somewhat similar thing when using both of them on my old monitor too, as soon as I enable color management for either of them they make gradients look a lot uglier than photoshop/irfanview/MS photo viewer do.

I wish I could edit the topic title to "I'm getting some weird things on gradients from Photomechanic and FVP" since it is now misleading as it turns out.

ONCE AGAIN: I think is a Photomechanic/FastPicViewer and doesn't likely have anything to do with the NEC PA profiles (although FVP does react particularly nastily in one case with one NEC PA profile, but again I dont see the mess in photoshop). Not sure what I am doing wrong with these two programs.




I was playing around testing all sorts of things, including lots of gradients and I notice that both Photomechanic and FastPicViewer get can some pretty horrific looking types of banding when viewing gradients and they are pointed to the monitor profiles that MultiProfiler or Spectraview II make. Red gradients get particularly brutalized. Worst is when you tag the gradients as sRGB and then open the display is set to either an sRGB profile made by SVII or even worse then it is set to a Photo profile made by SVII. Some pretty nasty sudden changes in red tone/saturation etc. can be seen along the 8bit gradient. They also really heavily block up the darkest shades on a gray scale gradient.

When loaded into Photoshop, or the latest version of Irfanview with the latest plugins and color-management turned on all the way, with windows photo viewer the same gradients under the same conditions don't look half as bad. The other two programs mentioned above really make gradients anywhere from slightly to disastrously worse looking than these other programs when viewing the gradients tagged with profiles and using the monitor profile as the rendering target especially when image gamma does not match display gamma and a conversion needs to occur.

I'll have to check but I don't recall such radically different results before with those two programs compared to the others.
EDIT: actually it seems I do with other monitors and profiles too

Not sure what is going on. Perhaps FPV and PM are trying to treat the profiles as absolute and then shifting up and that someone is reacting very badly with the internal LUT??

(yes dont worry though if you have the same tonal response set on the display as the file format the gradient has been tagged with and view in photoshop you will see no banding in sRGB simulation mode, just a nice band free gradient; so the 14bit LUTS are working)


----------
Some other things I noticed:
Windows Picture Viewer appears to be only partially color-managed. Seemingly not at all when in full-screen or slideshow mode. In windowed mode at the very least it converts image profile to sRGB and I thought also largely to monitor profile too but it definitely does not take display tonal response curve setting into account. So if you set your monitor's gamma to 2.2 and then use WPV to view sRGB images they won't match the look they have in Photoshop even on an sRGB gamut display, the very dark tones will look different.

sRGB pictures actually don't look the same when viewing in non-color managed vs. color-managed programs on even perfectly targeted to sRGB gamut displays unless the display is set to the sRGB tonal response curve. Something like photoshop will raise up the depths if it notices the display is set to gamma 2.2 instead of sRGB, same for irfanview, photomechanic, fastpicviewer and firefox but not much else as far as I can see. So everyone using say Internet Explorer actually sees the contrast on the deep end a bit off unless they have one of the rather few monitors that has been calibrated to sRGB instead of gamma 2.2.
« Last Edit: July 04, 2010, 03:26:21 pm by LarryBaum »
Logged

WombatHorror

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 299
Photomechanic + FastPicViewer seem to act weirdly with NEC PA profiles
« Reply #1 on: July 04, 2010, 04:16:53 am »

NOTE: as it turns out it doesn't seem to be related to the NEC PA profiles at all, but to using the monitor profile of any monitor
Logged

WombatHorror

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 299
Photomechanic + FastPicViewer seem to act weirdly with NEC PA profiles
« Reply #2 on: July 04, 2010, 03:46:11 pm »

OK looking over the Fast Pic Viewer documentation it turns out there is a registry key you can change to make it use the WCS floating-point color engine instead of the 16bit fixed-point windows ICC system. Selecting that does help a lot, none of the gradients on the NEC look like a disaster now when using it. That said they still don't look as good as they do in Photoshop or color-managed pluginified Irfanview, where they can look perfect (although it seems that irfanview applies ZERO color-management to PNG files).

It seems a bit odd though. If you switch photoshop over to windows ICC system instead of adobe engine it doesn't become a disaster at all and look almost as good as using the Adobe engine.

Anyway I would HIGHLY recommend using the registry key to unlock the WCS engine in Fast Pic Viewer, without using that it can make a total disaster of fine gradients with some monitor/profile combinations.

It's still a little odd though that even with that on, which you'd think would make it do at least as well as photoshop, it still doesn't do things quite as smoothly (OTOH it doesn't have the sudden tone drop below level 16 that ifran and photoshop have). For instance tag the 8bit gradient as an AdobeRGB file (gamma 2.2) and set the NEC PA to gamma 2.2 and view in photoshop adobergb working space and the gradient looks absolutely 100% smooth, same for viewed in irfanview. But load it in FVP and it has a bit of banding, not horrible, but definitely there.

Maybe for some reason one set of programs is trying to use absolute blackpoint and the others relative? (toggling in CS5 i don't see BP setting change anything at all though)



I can't seem to get PhotoMechanic's mess of banding to go away though.
« Last Edit: July 05, 2010, 02:47:30 pm by LarryBaum »
Logged
Pages: [1]   Go Up