The Big Bad
Senior Member
software seems pretty good except that it doenst offer anything except a JEPG
I realy would like it more if it could do a TIF
I realy would like it more if it could do a TIF
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
RAW mode to prevent blown highlights ?
- Thread starter The Big Bad
- Start date
Tim Lozinski
Leading Member
It depends on the kind of overblow you have. If you have a heap of sky which is too white, due to clouds being present, to make out detail in then you have to decide whether you retake the shot darker (given you will see it immediately at time of shooting) and try to raise the shadow areas later, or to bag the overblown shot and clone in a new set of details in teh sky later. Of course you can layer two variably converted RAW files and blend through the overblown areas to get the shot more detail in teh highlights. People might not like to do this but it DOES work, especially for when large highlights are simply too bright (ie large flat clouds or shooting with a bright sun in the scene)
check out this one I balanced out...
The shadows were too dark if the sun was not overblown (its not GREAT but its a sun, what can you do) but balanced out it doesnt look TOO dark when I tweaked the two different layers.
As for average highlights, like water in a landscape its really hard to make it work better if you have a glut of overblown area. What you need to do is avoid such scenarios by exposing as close as you can. Where you have occasional blown out spots where detail is not too important, but its still annoying to have white-out, you can clone in THAT situation, but otherwise its a lot more work than a simple clone stamp tool a few times.
if you have examples of what you might LIKE to correct overblown highlights on, post one or two, let me have a go on them and we can see if you like the method. That is- if its a doable solution, it wont always work so easily, but thats par for the course
Ed Halley
Senior Member
In just 10min after looking at that website, I experimented with the GIMP and found a method that worked on a quicky no-tripod pair of test shots pretty darn well. I don't know what the actual five algorithms might be for that software, but here is the basic GIMP/Photoshop method I used:
- Opened two photos that were lite and dark.
- Move the lite to a new layer over the dark.
- Close the original lite image.
- Convert the dark's Value to the Selection.
+ Duplicate dark layer.
+ Desaturate duplicate layer.
+ Channel Red to Selection.
+ Delete duplicate layer.
- Activate lite layer.
- Clear selection.
Fine tuning to the initial images, and to the selection gamut, and then further processing on the final image would benefit the process, but without any such tuning, the results are quite passable:
Lite image (1 stop overexposed by meter):
Dark image (1 stop underexposed by meter):
Combined image:
I'm going to script-fu this method (similar to Photoshop's "actions").
--
[ e d @ h a l l e y . c c ]
John Sheehy
Forum Pro
You're not getting the most out of your RAW bits that way, though. You're only using about 11 bits out of the 12, effectively boosting the ISO.
Under-exposing to compensate for the highlights always increases the noise in the image. Why not use the extra stop in RAW, if it's there?
--
John
John Sheehy
Forum Pro
The correct answer is YES.
Canon RAW mode for the DSLRs only uses about 11 bits for its default dynamic range, but there is a full 12 bits of data recorded.
If you use the Canon converter, you can only get the extra highlights if you convert to linear mode; even adjusting the EC compensation in normal conversion to -2 doesn't recover any highlights.
With Capture One LE, you can use the EC slider to compensate, or curves. C1 is near-realtime, so you can just drag sliders until blown highlights disappear. You can also use a default film curve to drag some extra highlights in automatically.
--
John
John Sheehy
Forum Pro
What camera is this? The 10D RAW files most certainly have about an extra stop of highlights, with the very brightest parts of this extra range being greyscale (R=G=B).
I have done side-by-side comparisons of linear TIFFs and the embedded JPGs. The TIFFs have detail where the JPGs have washed out at 255,255,255.
You can see this quite dramatically if you use a "curve" that is a 128-cycle triangle wave. The rings stop at the edges of the JPG wash-out, and continue into the area in the linear TIFFs.
--
John
John Sheehy
Forum Pro
Look, I don't know about the other cameras, but the Canon 10D raw data (after the analog-to-digital conversion) and CRW files contain values from 0 to 4095. When the camera is set to JPEG, anything above about 2000 or so (I forgot the actual point, and I plan to test this more thoroughly when I get the right target tool) is 255 in the 8-bit file, and 255 also in a RAW conversion from the Canon software (in non-linear mode).
There ARE extra highlights in the RAW file. You do not get them in JPEG mode. You do not get them with standard CAnon RAW conversion. If you use the Canon software, you must save in linear mode. If you use C1, you can recover the highlights with any tool; contrast, EC, curves, etc.
--
John
John Sheehy
Forum Pro
What camera is your reference? 10D RAW files most certainly contain about 1 stop more dynamic headroom that washes out in JPG mode
or default Canon RAW conversion (and is not available in the File Viewer with the EC slider).
--
John
John Sheehy
Forum Pro
It makes a tremendous difference. JPG files throw away the upper half of the levels recorded by the camera. This is a hard, cold fact.
Perhaps setting the contrast to -2 will recover some of the highlights in a JPG as well, but by default, anything over about level 2000 out of 4095 is tossed in the JPG conversion (and non-linear Canon RAW coversion).
It would have made the camera so much better if there was a variable-gamma mode, like film, for JPG conversion (and Canon RAW).
--
John
The Big Bad
Senior Member
That shot looks pretty good, i dont follow you in your method however, what is GIMP ?
Is this a method that could be done in photoshop as well ?
The results seem impressive though, Id love if you could clarify for me
Is this a method that could be done in photoshop as well ?
The results seem impressive though, Id love if you could clarify for me
John Sack
Forum Enthusiast
Well, in my neighborhood (silicon valley), GIMP is
GNU Image Manipulation Program. But I don't know what that would bring to this that photoshop couldn't do. I don't understand "Value to Selection" either.
I agree though: I can't follow Ed's instructions, but the results are attractive for sure.
Please write more, Ed!
I also wondered about avoiding or mitigating the situation of blown highlights in the first place. I've wondered whether -- in the high-contrast situations I shoot, where the overloaded highlights are ususall reflections of the sun on concrete, etc. -- a polarizer would help reduce the contrast to something the camera could handle.
John
GNU Image Manipulation Program. But I don't know what that would bring to this that photoshop couldn't do. I don't understand "Value to Selection" either.
I agree though: I can't follow Ed's instructions, but the results are attractive for sure.
Please write more, Ed!
I also wondered about avoiding or mitigating the situation of blown highlights in the first place. I've wondered whether -- in the high-contrast situations I shoot, where the overloaded highlights are ususall reflections of the sun on concrete, etc. -- a polarizer would help reduce the contrast to something the camera could handle.
John
In just 10min after looking at that website, I experimented with
the GIMP and found a method that worked on a quicky no-tripod pair
of test shots pretty darn well. I don't know what the actual five
algorithms might be for that software, but here is the basic
GIMP/Photoshop method I used:
GIMP has no "Value to Selection" directly:
- Opened two photos that were lite and dark.
- Move the lite to a new layer over the dark.
- Close the original lite image.
- Convert the dark's Value to the Selection.
+ Duplicate dark layer.
+ Desaturate duplicate layer.
+ Channel Red to Selection.
+ Delete duplicate layer.
This cuts holes in the lite layer where it's too lite, and lets the
- Activate lite layer.
- Clear selection.
details in the brightest part of the dark image show through
instead.
Fine tuning to the initial images, and to the selection gamut, and
then further processing on the final image would benefit the
process, but without any such tuning, the results are quite
passable:
Lite image (1 stop overexposed by meter):
Dark image (1 stop underexposed by meter):
Combined image:
I'm going to script-fu this method (similar to Photoshop's "actions").
--
[ e d @ h a l l e y . c c ]
Larry Lueck
Senior Member
John, thanks for the pointer to using linear mode! That's what
I missed in my testing, using the default FVU settings...
It does indeed allow a stop to a stop-and-a-half of blown
highlights to be recovered.
I think my landscape testing had skies blown by more than
1.5 stops, thus even C1 couldn't show them (although the
false-blue color that C1 'painted in' still concerns me...).
This will now make me rethink my 'mostly-jpg' stance...
N
Negative287
Guest
It doesn't matter what mode you're shooting in, because by the time the image is recorded the highlight would already be blown...
Correct metering and low contrast are the only effective ways that I've found to control highlight clipping.
Correct metering and low contrast are the only effective ways that I've found to control highlight clipping.
Ed Halley
Senior Member
Yes, the GIMP is GNU Image Manipulation Program. It's an Open Source alternative to Adobe's Photoshop. It doesn't duplicate every feature identically, but the essential concepts are the same: layers, channels, filters, etc.
Photoshop could do the same method, but I forget some of the interface details. When I said "Value to Selection", I meant, do whatever it takes to make the selection equal to the layer's value or luminance. Sorry that I can't just tell you the Photoshop methods, as I don't use it anymore.
Why GIMP? Open Source software allows me to (1) use the software at no cost, (2) have the source code so I can make modifications or explore methods if I like, (3) give copies to anyone I want at any time.
Why not Photoshop? Besides the obvious inverse of "Why GIMP?" (i.e., PS is proprietary, closed source, and very costly), I specifically do not support or recommend Adobe products. I am a programmer by trade, so the following issue means more to me than a few hundred bucks. Adobe used the DMCA to incarcerate a foreign programmer, Dmitri Sklyarov, on criminal charges, for doing and speaking about what was perfectly legal, but otherwise embarrassing to Adobe, Inc. Dmitri is now released but I still hold Adobe responsible for abusing this broken law that the media giants paid Congress to pass. Corporations have no vested interest in the rights of the individual. I vote with my dollars.
--
[ e d @ h a l l e y . c c ]
Dominic.Chan
Veteran Member
That's not true, for the reasons already explained earlier in this thread. Just because the JPEG pictures shows blown highlight, it does not necessary mean that the recorded highlight was already blown.
No one would dispute that.
John Sack
Forum Enthusiast
John,
What technique(s) would you recommend in C1? Just -EC until the highlight warning disappears seems easiest/fastest. Alternatively, developing two exposures and combining them seems best when you have the time. But what else is quickest and/or best? (I'm still learning C1).
I'm especially interested in how I might use Linear conversion combined with other tools/approaches.
I was shooting a political event yesterday (campaign kick off; its the season), and had to keep running from inside the house to outside (sunny hot day here in Northern CA). I barely had time to reset the ISO, much less run -EC. Because I took about 150 shots for the candidate, I don't want to spend "quality time" with each one until she tells me which ones are promising. So I'm looking for both "quick and dirty" (or, as the Brits say, "cheep and cheerful") as well as "best of breed" solutions in C1.
John
What technique(s) would you recommend in C1? Just -EC until the highlight warning disappears seems easiest/fastest. Alternatively, developing two exposures and combining them seems best when you have the time. But what else is quickest and/or best? (I'm still learning C1).
I'm especially interested in how I might use Linear conversion combined with other tools/approaches.
I was shooting a political event yesterday (campaign kick off; its the season), and had to keep running from inside the house to outside (sunny hot day here in Northern CA). I barely had time to reset the ISO, much less run -EC. Because I took about 150 shots for the candidate, I don't want to spend "quality time" with each one until she tells me which ones are promising. So I'm looking for both "quick and dirty" (or, as the Brits say, "cheep and cheerful") as well as "best of breed" solutions in C1.
John
John Sheehy
Forum Pro
Nope. Canon RAW files contain less than 11 bits of RAW data in the dynamic range that falls below the clipping point for JPGs and default FileViewer conversion (even with negative EC in the conversion). The rest of the data can only be extracted in FV by converting to linear mode, or by using curves, contrast, or EC in a program like Capture One.
The clipping point of a RAW file is way above the clipping point of a JPG, or default Canon TIFF conversion.
--
John
N
Negative287
Guest
HIghlights are clipped at capture time... Now some may not be harsh and may just be on the edge of going but once it's gone there is no getting it back regardless of whether it is RAW or JPG.
N
Negative287
Guest
Wait, so you're saying they're using a completely different tone curve for RAW files? Hmm, I wasn't aware they did that and just assumed they made them all equal so all things being equal once the data is clipped, it's gone.
Keyboard shortcuts
- f
- Forum