Here's some guessing on my part. Also, I explored this a while ago so some of my thinking may be a bit foggy.
1. I think that jpgs coming out of canon camera's emulate RIT.
2. When there was no aRGB, in RIT canon was probably attempting to keep the sRGB color space from looking flat by adding more white!?
3. Was DPP an attempt to generate better color after the introduction of aRGB?
4. If I was canon, I would not break RIT, I would add DPP to the mix giving photographers additional processing capabilities and let them decide.
5. If you are canon, how do you explain all this shit?
6. It takes some experimenting to figure this stuff out...
Thanks for the very interesting examples. We see the same trend as usual, DPP is killing detail in the highlights, as we can see clearly in the white feathers. Could you please let us know the settings used in RIT and DPP and which camera is it?
Sorry to revive an old thread, but since it seems that Canon has discontinued RIT and replaced it completely with DPP, I really can't wait to put my hands on a copy of the new DPP (v. 3.5) to see if they have improved the demosaicing algorithms and fixed the problems with highlight details.
Here's what I'm getting from the latest versions of DPP and RIT.
[DPP 3.4.1.1 and RIT 3.3 (3,3,0,5)]
[By the way, RIT is still available. It has never been discontinued, to
my knowledge.]
The shot, followed by 100% crops from the 8-bit TIFF files produced
by DPP (first) and RIT (second). These TIFFs were produced without any
modifications from the in-camera settings. The standard picture style
was used. The crops and JPG conversion were done in Photoshop Elements 5.
In playing around with DPP, it does appear that the neutral picture style
will retain more detail, and that the standard picture style applies more
contrast, thus blowing out the bird's body feather detail. Sharpening
in DPP is also a bit of a mystery; someone on this board reported they
thought DPP was employing a different sharpening algorithm than the
traditional USM algorithm.
I do notice a loss of feather detail in the body of the bird when using
DPP, but the eye of the bird looks sharper with DPP.
It would be interesting to see how the in-camera JPG file would compare,
but I didn't shoot RAW+JPG on this shoot. This was shot with a 30D
and a 400/5.6L, wide open.
Nice example, Tim, that clearly illustrate the point. IMO, DPP looks sharper because of it's demosaicing algorithms that seem to improve slightly OOF areas, but the loss of detail in the highlights is pretty obvious compared to RIT.
BTW, RIT has been discontinued in the sense that development has stopped and it will not be included in the future versions of Zoombrowser. Canon opted for 1 raw converter and I hope this issue has been dealt with in next DPP version.
An interesting excerpt from RG preview of the 50D.
"Note that RAW Image Task, the application that for some time has been bundled with ImageBrowser/ZoomBrowser EX and has handled RAW conversion duties for these programs, is no longer being updated. It will not be able to convert CR2s shot with the 50D or, presumably, subsequent Canon digital SLRs, though it will continue to be compatible with CR2s from most Canon models prior to the 50D.
For 50D shooters, Digital Photo Professional will be the lone Canon-provided RAW conversion option. Photos can be passed from ImageBrowser/ZoomBrowser EX to Digital Photo Professional for processing. There will be little reason though, in most 50D pro workflows we can think of, to use ImageBrowser/ZoomBrowser EX in the future. When shooting Canon, we've used ImageBrowser/ZoomBrowser EX really only to view AF point placement (a feature now slated for the next version of Digital Photo Professional) and to get to RAW Image Task.
Previously, RAW Image Task's RAW conversions were nearly an exact match for in-camera conversions to JPEG (except that RAW Image Task offers TIFF as an output file format). Assuming a low-compression in-camera JPEG setting, the colour, detail, noise, character of noise reduction and all other image quality traits were the same as they could be in JPEGs from a Canon digital SLR and a CR2 run through RAW Image Task.
Digital Photo Professional, by comparison, has offered conversions with finer detail and colour that is similar and pleasing (with the newest cameras' CR2s) or colour that is different and almost always inferior to RAW Image Task (with older cameras' CR2s).
As Canon fades away RAW Image Task, it will be interesting to see what, if anything, they've done to meld the look of RAW Image Task and Digital Photo Professional conversions together. Not to mention whether the 50D's DIGIC 4 processor performs in-camera conversions with the same level of detail as Digital Photo Professional."
I use ZoomBrowser EX all the time to view images. It has a nice feature that
shows the EXIF data along the right hand side of the window as you quickly
jump from one photo to the next, unlike the clunky DPP, which requires
ctrl-I to bring up a dialogue box for each pic separately. sigh..........
However, ZoomBrowser never could read the edit recipies from DPP,
so all the edits I made in DPP couldn't be seen when viewing CR2 files in
ZoomBrowser. Canon has never been too serious about getting their
tools streamlined and working together well.
Hopefully the new DPP will be better in every way. Oh please, give us arbitrary
image rotation angle and simple cropping. Why should we have to export to
Photoshop just to straighten horizons? heavy sigh..........
You should be able to devise your own picture styles, using Picture Style Editor, which will handle highlights however you want. Yeah, it'd be more work than just switching Raw converters.
Off-topic:
trumpet_guy wrote:
Oh please, give us arbitrary image rotation angle and simple cropping.
DPP does cropping.
therock wrote:
I have trouble getting to RIT. In my program files I have an uninstall and a readme thing but no exe.
It's not a separate program. It's part of ZoomBrowser EX, or if you have a Mac it's part of ImageBrowser.
Yes I have found this to be true. RIT is designed to give a sharper more ready file with less processin and is basically identical to the camera JPEG you would have obtained if shooting in JPEG mode.
Aside from zoombrowser's better details, it would be nice if the new DPP can have delete-file button while in Edit (raw) Image mode so I can delete OOF shots - with this, Zoombrowser will become unnecessary.
trumpet_guy wrote:
Sharpening in DPP is also a bit of a mystery; someone on this board reported they
thought DPP was employing a different sharpening algorithm than the
traditional USM algorithm.
About a year ago I posted something comparing two versions of DPP (3.0.1 vs. 3.2). The thing that caught my eye was how the two versions each had different sharpening characteristics for each channel (RGB), but more so the red and blue channels. In one the red channel was sharper than the other while in the other the blue channel characteristics were flipped. The specific image I compared had areas of highly saturated yellow and each conversion showed much different sharpness/detail:
Why exactly this was done, I'm not sure. Maybe to suppress blue channel noise, which is generally the worst of the three. I reprocessed that image through DPP 3.4.1 and there is no significant difference in the red channel, but it does appear that the blue channel, while still looking like it's been blurred, has better edge sharpness than v.3.2.
Anyway, Canon must have a reason for this. Maybe DPP is optimized for certain applications, like people photos where skin tones are critical. A sharp blue channel might accent skin blemishes more, and the kind of highlight definition shown by RIT might not be considered as pleasing.
While I can live with the sharpness characteristics of DPP's highlights (maybe because I don't do bird photography), I'd really like to see a dedicated highlight slider that would allow better definition/recovery of highlight tones when desired.
trumpet_guy wrote:
Here's what I'm getting from the latest versions of DPP and RIT.
[DPP 3.4.1.1 and RIT 3.3 (3,3,0,5)]
[By the way, RIT is still available. It has never been discontinued, to
my knowledge.]
The shot, followed by 100% crops from the 8-bit TIFF files produced
by DPP (first) and RIT (second). These TIFFs were produced without any
modifications from the in-camera settings. The standard picture style
was used. The crops and JPG conversion were done in Photoshop Elements 5.
In playing around with DPP, it does appear that the neutral picture style
will retain more detail, and that the standard picture style applies more
contrast, thus blowing out the bird's body feather detail. Sharpening
in DPP is also a bit of a mystery; someone on this board reported they
thought DPP was employing a different sharpening algorithm than the
traditional USM algorithm.
I do notice a loss of feather detail in the body of the bird when using
DPP, but the eye of the bird looks sharper with DPP.
It would be interesting to see how the in-camera JPG file would compare,
but I didn't shoot RAW+JPG on this shoot. This was shot with a 30D
and a 400/5.6L, wide open.
Edited by trumpet_guy on Sep 05, 2008 at 02:25 AM GMT
Edited by trumpet_guy on Sep 05, 2008 at 02:29 AM GMT...Show more →
I have quoted the above to ease the comparision with the latest results below.
Now that I have DPP 3.5.1.0 installed, I have reconverted this picture from
the RAW file using this latest DPP version. The conversion was done just
as before, with default settings in Standard picture style (which is +3 on the
sharpening). An 8-bit TIFF file was first made and opened in Photoshop
Elements 5.0 then a 100% crop was made and converted to JPG using
Save As with the highest quality setting (12).
It seems like the algorithms in DPP have once again been tweaked.
To my eye, this latest conversion shows less sharpening and overall "pop"
than either RIT or the previous version of DPP. Maybe I just need to apply more
sharpening in DPP now. To me though, this latest change is a step in the
wrong direction.
I say that Canon instead of improving and developing RIT further has discontinued it. We were all hoping that the new DPP will do similar conversions to RIT but .....
It's discouraging to see the highlight detail reduced.
I've not found a way to pull that detail out of the feathers using this
latest DPP.
One would presume that Canon wouldn't mess with this part of the demosaicing
algorithm unless they are improving things. I don't get it. If it's not broken,
don't fix it.
