Вопрос:
Quote from: DustinFrancis on March 03, 2018, 04:01:04 PM писал(а):
I have a question I am wondering if anyone can answer. I have a film scan of a Color Checker SG that I measure the patches of in 3DLUT Creator and then use that data with a Babel Color Patch Tool reference file to create a DNG Profile of in Lumariver Profile Designer. When I render the profile, the patched definitely change to close to the reference patches...especially the hue, but almost all of them are quite a bit more saturated and either lighter or darker than the reference patches. I have everything in the Profile Designer set to default and have tried utilizing almost all of the tools and options available, but that only makes things less accurate/pleasing. My main question is, how do I get the rendered profile patches to match the reference file patches?
Ответ от разработчика DCamProf:
* Выделение мое.Today at 02:58:47 AM писал(а):
I'm not sure what you are trying to do here, but it seems like you are not trying to make a film simulation profile? That is you use a film scan to make a reference not of the real colorimetric colors, but the film response, and then use that file with a target shot of a regular camera, in order to make a profile that makes that camera look like the film? That's probably not going to work. Lumariver Profile Designer is not designed to be a generic LUT Creator software, but is designed to make camera profiles anchored in realism with only smaller subjective elements. It makes assumptions of how camera/scanner hardware works. This means that it always starts off with a linear matrix and make LUT corrections on top, which make robust smooth profiles, but requires that the system profiled is at large linear. If the behavior of the camera (or the reference file) would be grossly non-linear, that is the camera/reference does not work as real cameras do, the matrix matching will go haywire, and then the LUT correction, which is supposed to make only mild smooth refinements on top, will get into trouble. I haven't really tested these workflows but I wouldn't be surprised if you get crazy results or even if the software fails in optimization and get stuck.
For creating film simulation LUTs there are better tools out there, Lumariver Profile Designer is not intended for that purpose. It would require a different design which does not use a matching matrix at all but instead work with really dense targets (lots and lots of patches to cover the full gamut) and make a complete non-linear match.
Хотя все это и так было понятно и расписано давным-давно и подробно у него здесь: DCamProf manual
Так что я настаиваю, что мечтать о построенных по колорчекерам профилям неземной красоты лучше бы не надо.
Ну можно еще помечтать о том, что Вы сами "на солнце и в тени/под линзу/исо" построите _просто_технически_реалистичный_ DCP лучше, чем, к примеру, дядьки из Adobe в своих лабораториях.
Но простая проверка (о которой написал выше) не подтвердила такую возможность.
У меня вообще есть глубокое подозрение, основанное лишь на _теории_, что Single-Illuminant DCP - это, как просто шампунь, в то время, как Dual-Illuminant DCP - это шампунь с кондиционером.
Тем не менее в X-Rite ColorChecker Passport есть _отличная_ карта для баланса белого, жаль только маленькая.
(Только меньше курите травы, присланной из Голландии).
Да, а тема действительно старая. Я с Юлей зашел поздороваться, почему сюда, даже не знаю. Изв., если сбил темп. А с самим профилированием давно все более или менее ясно.