replicating Oly colors in LR... impossible?

Started Jul 22, 2012 | Discussions thread
gollywop Veteran Member • Posts: 8,284
Re: replicating Oly colors in LR... impossible?

assaft wrote:
Hi gollywop,

Thanks for the guide!

In the guide you say this -

You can make different targets for different sunlights (bright daylight, cloudy, overcast, shade). I've done this but find that a well-made target for typical daylight (5000-5300 °K) makes an excellent overall profile.

And previously on this thread you said:

My interest is in beginning with an image whose colors are appropriately close to the originals. I do this using custom camera profiles, shooting raw, and using a WhiBal card to set WB.

I'm trying to understand the workflow -

In the field you ETTR using UniWB, and then in ACR you apply a general color profile (created primarily for 5000-5300K but used under any light condition) and you override the default WB of the profile/UniWB based on a snap of the WhiBal?

That's pretty much it. I develop the image in ACR, using sRGB for web-destined images and ProPhoto for printer-destined images. If I like what I've got, I go into PS only for final resizing/resampling and output sharpening. If I feel it needs yet more, I do further post-processing in PS. I used to do this in LAB, but now I stay in RGB, particularly if the final image is going to remain in sRGB for the web.

btw, as far as I know color profiles improve the accuracy of the raw converter's color reproduction, but they do not take the manufacture jpeg engine into account at all. So if the Oly jpeg colors are hard to achieve because they are tweaked and unnatural (e.g. due to some boosted reds), a color profile won't be able to match them (and is not supposed to). I wonder if there is a tool that can learn the color tweaking automatically based on examples. I mean, a tool that will be trained on pairs of RAW+JPEG (created by the manufacture (Oly) jpeg engine), and then will generate an ACR color profile that produces the closest color rendering when converting future RAW files (based on statistical machine learning methods). My feeling is that if the color tweaking in the jpeg engine is static then it is feasible to develop such tool; if it is dynamic (e.g, based on recognition of the type of scene) it would be quite complicated.

I know of no such application. I guess I wouldn't use it if it existed. I don't find the Olympus colors to be particularly special, and they certainly aren't particularly accurate. Many of the camera manufacturers appear to deliberately pump colors because they've found that customer's like a warmer this or cooler that, yellower greens or more alabaster (refined) skin.

As I recall my earlier OOC jpeg vs. raw comparisons, Nikon (at least in the pro/semipro models) tended to be least offensive this way. I can't say anything about the top model Canons since I don't have one, but it's sure true of the s60 or s90. I guess I feel that, if I want to pump colors away from being accurate, I want to do it according to my taste, not some marketing guy's. Perhaps I'm being unfair, but it doesn't really matter to me. I want to do my own CM anyway. All of my cameras, using my custom profiles either in ACR or RPP, produce beautifully consistent results.

Thanks again,

You're most welcome.

-- hide signature --



Post (hide subjects) Posted by
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow