Colour loss, raw to jpg conversion

what built-in tools? as Rob said, which applications? i don't care what images look like in non-color-aware applications at intermediate stages until i am at final stage and producing JPGs for the web. then i care that the sRGB low resolution ones look as similar as possible to my ProPhoto RGB working files in Photoshop. with a calibrated monitor, when everything is in sRGB, even non-color-aware applications should look identical.

Herb....
 
Yes, the two files are : original raw, viewed in 'RawShooterEssentials', the other file the converted jpg, viewed in Photoshop Elements. Note that i have tried conversion using the Pentax software that came with the camera, with roughly the same result.

Comfort: No matter what the reason turns out to be, I have the raw files
;.)
--
Kurt S.
 
Perhaps it would be a good idea to invest in a calibrator ?
f.ex. 'Pantone Huey Monitor Calibrator'

does anyone here use it ?
--
Kurt S.
 
This is what I usually expect between RSE and conversion.
RSE to sRGB jpg. Opened in Corel w/ NO color management.



--
360 minutes from the prime meridian. (-5375min, 3.55sec) 1093' above sea level.
 
Raw gives you a better picture, with more detail, and more color differentiation. But, like you say, raw opens up a ton of new variables.

I used to use RSE with a Nikon D70. It never got the color right. Mostly it seemed to be the White Balance. I gave up.

I've used a Pentax DL with Photoshop Elements 4.0. I stay away from the 'default' setting, which includes a lot of 'brightness adjustment' and compensation 'exposure compensation'. Usually the basic color, the White Balance, is fine.

If you use a three color 'level' you can usually get the color you want. Sometimes the 'color cast' correction will clean things up.

With raw, you have a billion variables. Generally, most people can avoid calibration if they use on monitor and a printer. You just get a feel for the color. If the program you are using 'never' gets the color right, you have to try another program. I think.

I use an Epson 1270 and a Canon 6000. I can usually get what I want, but sometimes I use different amounts of saturation.

"Workflow" is not a good concept until you know enough to see how the colors are wrong. Is the problem White Balance, or calibration, for example? Is the color 'wrong' on the monitor, or moving from monitor to printer? For the most part, if you print in a program, work from that color and don't worry that things look different in the first step.

In other words, if RSE output looks different in Elements, work from Elements to your printer or Web output. If you can't do that, because RSE screws things up, ditch RSE.

Just some thoughts....

George Sears
 
Hi,
I seem to have solved the problem, or at leat found the explanation :
I had calibrated my monitor using Adobe Gamma.

Now, if i set gamma to 'Windows default' insted of the 'custom' setting I have been using lately, there is no longer any marked difference between raw and jpg!!!

So far so good, but i don't understand why.... surely, the monitor should be using the same profile, no matter what program i am in????

I am considering 'Pantone Huey Monitor Calibrator' anyway - but would it be a waste of money..? I wonder....?

--
Kurt S.
 
I'd work with it a while, see if you can get a handle on the color. Is it a white balance thing? Is there a shift to red or green. Just follow the sequence from where you start, raw conversion, to printing.

When you have a real idea of how much something is 'off' then you can try to very precise calibration, if you need to.

Spend some money on ink, if you have to, first. Generally, most people can figure out what they are going to get and make small adjustments.

The real problem is when a program just can't get a good white balance. Then a calibration won't help. And you generally can't adjust it, white balance, after the raw conversion is done. So if the raw converter is giving you something 'muddy'....

Remember you can do color levels, hue adjustment, saturation adjustment, and color cast adjustments. Too many things.

You really want it pretty close to right after raw converstion and maybe one color adjustment, one tweak like levelling with all three colors.

Hope this helps. Glad you're getting it sorted out.

George Sears
 
I seem to have solved the problem, or at leat found the explanation :
I had calibrated my monitor using Adobe Gamma.
Now, if i set gamma to 'Windows default' insted of the 'custom'
setting I have been using lately, there is no longer any marked
difference between raw and jpg!!!

So far so good, but i don't understand why.... surely, the monitor
should be using the same profile, no matter what program i am in????
Ah, if only! Here's some advice which will cost nothing to try. I warn you this is not in agreement with expert opinion, so I'll just add that it's intended as a "quick fix" for those who mix'n'match applications that are colour-managed with those that aren't, and addittionally they don't have a fully colour-managed workflow... which sounds like you ;-)

Calibrating a monitor with Adobe Gamma (or other similar app) does two things: it creates a LUT-loader to adjust your video card at boot time - this affects all software. But it also creates a crude ICC profile for software that bothers to use it. Therein lies the problem - not all software uses it, hence the differences you see. If you get rid of Adobe Gamma you lose the correction to the LUT and the profile, so the colour will be equally bad everywhere ;-)

But there's another way - keep the corrected video LUT provided by Adobe Gamma but get rid of the ICC profile generated at the same time.

In "Display Properties" click "Advanced" and select the "Color management" tab. You'll find the profile that AG produced selected as the default. Remove it, leaving no default.

That's it. Everything will work fine, screen colour should be consistent across all applications whether or not they are colour managed. This affects only display colour management, it won't harm other colour management in apps that have it.

If you don't like the result just run Adobe Gamma again and re-save the ICC profile.

--
John Bean

PAW 2006 Week 12:



Iindex page: http://waterfoot.smugmug.com
Latest walkabout (6 March): http://waterfoot.smugmug.com/gallery/1259425
 
Hi John,

Thanks, that makes sense to me... it would seem, then, that RawShooter ignores that other ICC file, whereas Photoshop Elements use it ? That would explain everything.

Actually, i have been considering 'upgrading' to CS2, and perhaps buying the 'full' RawShooter packet, which I understand has new Pentax colour profiles? Does anyone have experiences with them?

--
Kurt S.
 
I have tried RSE and did not like it at all. I could eek out a little more detail but the color was always off. The latest version of ACR gets the color right and tends to start out close to where you want to end up. I tried a version 2.something and it was good but required a lot more work than the newest version.

I now have PS CS2 and there are some tools that can get a JPEG to look a lot like a RAW image. However, you will end up noticing the limits of JPEG dynamic range. I am also trying +1 sharpening and -1 contrast on my JPEG's to see if I get better results.
 
Well, my problem was : the colours in RSE were exactly as I perceived them, but the moment I converted to jpg and viewed in AE3, the colour had changed, though other facters like sharpness, contrast etc, were ok.

But, as you see in my answer to John, that all changed when i reset gamma in Adobe Gamma to Windows default.
--
Kurt S.
 
Shooting JPG, I notice "corect" colour in CS but ALL other applications have a slightly flatter tone and a definite magenta cast. I can view this in CS by selecting "Proof setup> Monitor RGB", whilst this offers consistancy the colour is incorrect .... so it is tough to choose which to go with.
--
***********************************************
Please visit my gallery at http://www.pbase.com/alfisti
 
Kurt: RSE/RSP are aware of colour profiles. You just have to tell the application about what profile you want it to use.

Unfortunately RSP doesn't handle colours from the DS right. Red gets orange. The extra colour profiles needed are in the addon package called color engine. The total is more than SilkyPix. Your choice...

Pixmantec seems not to be very communicative. I mailed M Tapas a week ago and again yesterday about this thread here:
http://forums.dpreview.com/forums/read.asp?forum=1036&message=17671359

I showed up there with some text advertising for RSP. He still haven't replied to my mail though. Maybe I should sell my RSP license while saving money for SilkyPix....

George: I didn't know "people" hated shooting raw? Are you sure?

Jonas
 
Well, as far as i know, Pixmantec is a danish company, so, being a dane, perhaps I should ask them in plain danish??? You know, 'hvad helvede er meningen?* ? ;-)
--
Kurt S.
 
Ja for faen.

Go right ahead. I didn't say in my mail to M Tapas that I'm going to publish the mail, so I won't. However basically it linked to the post I linked to above. Then I asked Tapas why my "fully featured", and paid for, converter is unable to produce red colours from my DS (which is said to be supported). A good colour profile should be included in the full featured product... Is there anything that can be done, like a free DS profile perhaps? That's about it.

til lykke,

Jonas
Well, as far as i know, Pixmantec is a danish company, so, being a
dane, perhaps I should ask them in plain danish??? You know, 'hvad
helvede er meningen?* ? ;-)
--
Kurt S.
 
has been getting the contrast settings correct in my experience. RSP performs much better if you start with a strong midtone adjustment (1.3 to 1.4) and reduce high/low contrasts. These types of adjustments do result in and have cured some color shifts.

I've concluded that perceived accuracy of colors depends as much if not more on the smooth transition between colors rather than the absolute accuracy of any individual color. That's not to say RSP ColorEngine and other profiles aren't worth the money.

White balance hasn't been a problem with any of the RAW converters I've tried as long as there is something white/gray in the frame or I use manual WB in the camera. Auto WB is somewhat hit or miss for all converters (including the one in the camera) I've used. Mixed lighting is especially difficult.

RG
 
I owned a Minolta A1 a while back. Some coverters were so far off on the color it was hard to believe the product would have been offered to the public. Adam T has done tons of work with converters with the Minolta A1, and he seems to gripe about the white balance on RSE. There were complaints on the Nikon SLR forum about RSE and general complaints about color.

Most of us will download something, but if it isn't successful in a few minutes, we might give up. RSE has interesting settings to bring out detail that some people like. I've seen a lot of feedback even on the RSE site about certain profiles, changed profiles, but I gave up.

How far do people want to go? When people start using 4 different converters, it gets a little ridiculous. The jpg you get from the camera is a good starting place. When you use the basic Nikon converter, you get more or less THE NIKON jpg, but with more detail and better color. When you go to RSE, and convert, you don't. Most people want the raw conversion to look like the jpg, or be better than the jpg. But a lot of people find the raw conversion has worse color than the jpg, or they are conditioned to the jpg.

Most people who start using raw expect it to look like the jpg, but with more detail and more exact color, smoother color transitions.

If you use a converter and adjust midtones, you are definitely changing the curve. How much of a curve does the camera put on the jpg? A Nikon raw converter may put that curve on the raw output, but others may not be able to get it exactly right. Raw introduces a ton of variables. It doesn't seem very exact if you have to do a midtone adjustment, but if you do it everytime, it might be easy enough.

What you are saying about smooth transitions of tones is interesting. JPG has limited tones, limited color, so transistions will be less smooth, right? So if someone says a jpg has better color than a raw pic, does it go against your theory? How many ways can a raw converter screw up the color? What assumptions are they making? Copy? Improve? Make a bold statement?

Assume you take a perfectly exposed picture with easy white balance, balanced so you get a nice even histogram, edge to edge. Assume the jpg is flawless, with perfect color, a perfect histogram for every color, great whites, whatever. Now, if I open that with a raw converter, and use whatever the program says is default, should everything still be perfect? What should be different between the jpg, from the camera, and say a TIFF from the raw converter?

George Sears
 
Didn't I mention that going down that road was pointless???? :) Maybe after I discussed it w/ Michael long before RSP and long before current RSE that they completely dropped the issue. Nice that you tried. I don't think they think it's broken just different and most of the times they think it's a wb problem......... sigh.
Just good to know some converters do it right
--
360 minutes from the prime meridian. (-5375min, 3.55sec) 1093' above sea level.
 

Keyboard shortcuts

Back
Top