| Author | Thread |
|
|
09/15/2007 03:18:51 AM · #26 |
Thanks Colette. Yes, Lightroom is definitely extracting extra highlight information as well. Good to know.
|
|
|
|
09/17/2007 09:09:32 PM · #27 |
Bump - nobody got any samples from DPP? I've tried a few other free softwares, DPP, Faststone, Picasa, Rawshooter Essentials. All can do the basic conversion from RAW to JPG, but none of them are able to extract any more information than the JPG. It's like they do the 8 bit conversion first, and then do their adjustments.
I also tried BreezeBrowser and CaptureOne demo versions, and both seemed to work well, getting that little extra information out of the RAW highlights. And ACR and Lightroom are shown in this thread to be useful.
Anyone got anything useful from the free softwares?
|
|
|
|
09/17/2007 09:12:10 PM · #28 |
Are you applying exposure adjustment before rendering/exporting the jpeg? Or are you converting to jpeg using defaults and then looking for any differences?
For the test I did, I made only one major change, and that was to set exposure to -4. I also adjusted brightness and contrast, but that's just because I wanted to level the playing field in case further tests needed to be made.
Message edited by author 2007-09-17 21:12:36.
|
|
|
|
09/17/2007 09:45:16 PM · #29 |
Yes, I'm adjusting the curves/exposure/etc in RAW before converting/exporting to the JPG. :)
|
|
|
|
09/17/2007 09:58:22 PM · #30 |
| Why don't you send a mail or put a call into canon support, give them your opinion on the software and then ask if it's really that shitty or if you are missing something? |
|
|
|
09/17/2007 11:28:55 PM · #31 |
Originally posted by surfdabbler:
Thanks for the sample image,  viajero. Yes, on this image, I can see the benefit, so I'm glad to see it. There is certainly more detail in the converted RAW compared to the JPG. This is something I just cannot get from my RAW images adjusted with DPP.
I'm interested to know whether this is standard image settings in camera (contrast/colour/etc). Also, what software did you use for the RAW conversion? I'm happy to post raw images if someone wants to try to get more info out. I will download the latest update of DPP, and have another crack at my images, and if I still can't get anything, I'll put the RAW file on my server, and see if anyone else wants to show me what can be done with different software. |
Recently I've been using the RAW opener that's built into CS2. I mess with all those settings, with most of the effect coming from the color balance, exposure slider, and the adjustment curve.
Before I had CS2 I used Elements and the RAW conversion software that came with the 20D; I forget what it was called. I would often do two conversions, one using the default settings and one using the "linear" mode. I would paste each of those files into a separate layer and blend them.
I forget which method I used with the goat picture, but it was one of the two described above.
My contrast, sharpness, and saturation were all set to +1 for that goat picture.
On the 20D those three settings default to 0 and can each be set to -2, -1, 0, +1, or +2. The meaning of these numbers varies from camera to camera. On the 1D series cameras 0 through 5 correspond to -2 through +2 on the 20D, with the default being 0 (-2 on the 20D). Canon point and shoots default to what would be +1 or +2 on the 20D, but of course they number it differently. Somewhere there's a Canon document that explains the differences between cameras.
It should be noted that these three things are what cause most of the "destruction" in going from RAW to JPEG. If you set all three to the minimum allowed, you would get a flat washed-out looking JPEG that would have kept as much detail as possible from the RAW file (but still not quite all of it). It would also require more post processing to make it look good, just like a RAW file would.
So, that's why I take the disk space hit and shoot RAW+JPEG. With the JPEG set to a more "point and shoot" type setting with more contrast, sharpness, and saturation, I get JPEGs that usually look good right out of the camera. Then if I really want to brighten some shadows and/or get all the detail out of the highlights, I can spend the time monkeying with the RAW file.
|
|
|
|
09/18/2007 12:01:53 AM · #32 |
Originally posted by routerguy666:
Why don't you send a mail or put a call into canon support, give them your opinion on the software and then ask if it's really that shitty or if you are missing something? |
Done. :) Thanks for the suggestion. I thought of this, but I don't know why I didn't do it before. Canon agreed that it should work, but couldn't explain anything more yet. They will "get back to me".
|
|
|
|
09/18/2007 12:19:40 AM · #33 |
I'll have a go with it. I was intending to do this a bit earlier, but I had the groom arriving on Monday at One PM... Finally sent him on his way at 6. Gotta love OS crashes...
I loaded the image into DPP, then dialed the exposure down using the brightness. I saw no change in the amount of information in the blown out highlights area when using the mouse hover method...
Then I thought, maybe this is the image representation. Not the actual information, just the pre-load JPG used by DPP to show the results.
So I processed it.
And checked again.
The results were the same.
If you look at the histogram, there is a cut-off where the original image tops out.
This leads me to believe that DPP is looking at the image with 8 bit eyes. Very strange. It's almost like the image is processed to JPG before the conversions are effected.
I haven't done lots and lots of stuff with DPP, (it's primarily my first pass sorting tool and white balance correction), but I do use it. I'm not too happy about this discovery and I am not really satisfied.
I will be looking into Lightroom and ACR.
Surfdabbler, I will add my voice to yours if you want to PM me with some information as to how you contacted Canon. I am not local to the US, so going the normal route wouldn't do much.
Message edited by author 2007-09-18 00:21:12. |
|
|
|
09/18/2007 12:34:43 AM · #34 |
Well, I've actually manged to process the file in ZoomBrowser, using the RAW Image Task, and I can see some extra RAW information there, so I don't think the whole DPP thing is a deliberate effort on the part of Canon to restrict their free software. :)
I'm in Australia, so I just called the local support number in Aus. We'll see what they come back with.
|
|
|
|
09/18/2007 05:00:15 AM · #35 |
I got mucking about with Adobe Bridge which I hadn't used previously due to not having a zoom feature.
Lots of really excellent features in the Bridge. The fact that there is no loupe tool or zoom in their full screen mode is MADDENING. I really enjoyed that part of the CS3 demo.
I believe that I will now use DPP for sorting only, using the 1,2 and 3 star method with simple folders for quality hierarchies. I will then work on edits and processing to RAW via Bridge. With ACR, it really is quite excellent.
I am very disappointed in DPP.
thanks for pushing the subject surfdabbler.
Now here's hoping my friend doesn't ask me to give his CS2 back... heh. |
|
|
|
09/18/2007 05:19:47 AM · #36 |
Originally posted by eschelar:
This leads me to believe that DPP is looking at the image with 8 bit eyes. Very strange. It's almost like the image is processed to JPG before the conversions are effected. |
The RAW plug-in which comes with (free) Nikon Picture Project also converts the RAW to an 8-bit image. Which of course totally defeats the purpose of shooting RAW in the first place.
Interestingly enough, the RAW plug-in which comes with the ($120) Nikon Capture NX works in 16-bit. |
|
|
|
09/18/2007 06:22:09 AM · #37 |
?
am I reading you correctly?
Every RAW converter converts to an 8 bit image. What we are talking about in this thread is the possibility that the converter is converting it to an 8 bit image BEFORE applying the adjustments that are supposed to make RAW better.
In the event that we have some crossed wires, I will explain what I mean.
8 Bit images are images which have 8 bits to express each color channel for each pixel. 8 bits for red, 8 bits for green, 8 bits for blue. Because these are luminous pixels when talking in the context of a computer screen, these 8 bits can create full color images that we are used to seeing.
When you are viewing things on your computer, for the most part, your computer screen will only be able to view in 24 bits, but this is mimicking 32 bits of CMYK. The K is not necessary on a computer screen, so it is used in other ways. see here
working with images in higher information color spaces, such as tiff and DNG or whatever really is difficult to accurately determine with a monitor or LCD screen, and for these purposes is largely pointless. These only become really useful when you are going to print fine art or something where the printer itself is capable of handling deeper color depth.
To this end, virtually all RAW converters can also convert to 16 bit tiff and the like.
When working with a deeper color depth, gradients are smoother, particularly in areas of very low contrast (flower petals, brides dresses, certain sepia and B&W images...).
For this reason, few people find a need for a final product with more than 8 bits per channel. As near as I can tell, even with my funky 9600x2400DPI 4 color printer, 8 bit is probably the working color depth.
In the context of RAW and JPG, a JPG is a small box which holds a certain amount of information. RAW holds more information, but still represents the same color RANGE on paper (see color theory and zone theory - black always represents black and white always represents white = the bounds of paper). Therefore, converting from RAW to JPG means throwing information away. A RAW editor allows you to CHOOSE the information that is thrown away.
Shooting JPG in camera generally chooses the center of the camera's range, squeezing it a bit to fit according to the choices made by you in the camera settings and by the manufacturer. Once you get to the level of JPG, any adjustments you make to information internal to the JPG run the risk of overlapping or 'information gaps'. Overlapping can cause clipping and loss of detail in low contrast areas and 'information gaps' can be seen in your histogram as gaps between the flowing 'peaks and valleys'. Push hard enough and you will throw away enough internal information that you will lose realism.
What is being talked about in this thread is moving the choices towards one end of the other of the total information range gathered by the camera's eye (12 bits per channel) and choosing to throw away the part of the RANGE where little or no information is present. For example, either the top 8 bits or the bottom 8 bits.
However, what we are finding is that some RAW software appears to have thrown away the top and bottom ends of the information, so when the box is moved in a particular direction, nothing new is added. You can see this effect in the Histogram as you adjust exposure in the various different RAW editors mentioned above.
If you move the exposure slider and there appears to be a gap at the end, then the effect is that the RAW converter is looking at the RAW image with '8 bit eyes' and is unable to see the extra information.
If this really is the case, then there is no benefit to shooting RAW at all. It has been demonstrated that Adobe can see more information, so this is not really the case.
Pardon my lengthy explanations. I've been doing a lot of reading all over the place and it seems that not everyone is on the same page with these definitions.
/jack
Point is, converting TO 8 bit JPG's is not totally worthless, but is different from working IN a 16 bit space.
It may just be that the programs themselves have become limited to 8 bit color spaces. Rather strange. I have just converted the above RAW file to 16 bit .tif and I am still not retrieving any more information. This is counterintuitive to what should be happening, where the 12 bit information should actually be getting stretched to 16 bit and nothing should be thrown away. DPP is still giving me information in the 8 bit range and is still unable to reveal any more information in the 'blown out' areas via manipulation with levels or curves.
Message edited by author 2007-09-18 06:31:18. |
|
|
|
09/18/2007 06:37:06 AM · #38 |
Yes, that's a pretty clear and concise explanation, I'm in agreement with most of it. Apart from this piece;
Originally posted by eschelar:
Every RAW converter converts to an 8 bit image. What we are talking about in this thread is the possibility that the converter is converting it to an 8 bit image BEFORE applying the adjustments that are supposed to make RAW better. |
A good RAW converter should be able to convert to a 16-bit image. This would a) retain the 12/14 bits of colour information per pixel and b) allow 16-bit image editing in PS. Basically this would allow you to do all the 'exposure' adjustments in PS that you would otherwise have to do using your RAW software (since that's all your RAW software actually does; convert the RAW data to a 16-bit image and allow you to edit it in 16-bit mode)
As you pointed out, if your RAW software converts to an 8-bit image, you've just lost the only advantage to shooting RAW: The ability to edit the image using all 12/14 bits of per-pixel colour depth.
I'd love to know which bits get 'thrown away' when the conversion to 8-bits happens. Or does it compress the 16-bits into 8-bits?
|
|
|
|
09/18/2007 07:29:25 AM · #39 |
right. I'm not saying that since a RAW converter can convert to an 8 bit image that it cannot convert to a 16 bit image. Every RAW converter that I have used has several output options, some in 8 bit color spaces, others in 16.
Theoretically, converting from 12 to 16 bits would retain all of the color information. It is possible that this isn't the case though.
The bounds of the 8 bit JPG box are the same as the bounds of the 12 bit box and the same as the bounds of the 16 bit box when considering the BOTTOM END and the TOP END.
the 0 value is as low as it can get. And the FULL value is as high as it can get.
(Sooner or later, someone is going to point out that the primary difference between JPG/TIFF and RAW is that for each pixel location, RAW only has one color channel of information and the other part of processing is the interpolating of other channels into this location, but this is not relevant for this thread)
This is true because of the reference to the DISPLAY MEDIA. If using a computer screen, the full values are (USUALLY) R-255, G-255, B-255. This is full white.
A RAW converter will therefore do a preload to 8 bit (sometimes called an '8 bit JPG preload image' or something similar) to work with your monitor's display and these are PRESUMABLY the values which are returned when you are moving your mouse through an image. This is why the folks earlier in the thread made strong changes to the images to severely emphasize the subtle differences in the values of the 'blown out areas'.
When converting from 12 bits in RAW to 16 bits in Tif though, the information will need to be STRETCHED in order to make the bits of information at 0 correspond and the bits of information at FULL VALUE to correspond. This will introduce gaps in the actual information which can be nicely filled up by the converter's conversion algorithms.
On the other hand, when moving from 12 bits to 8 bit JPGs, you STILL NEED to make the information correspond. In the simplest of worlds, you would simply throw away the top two and bottom two bits. Visually this makes the image punchier with stronger blacks and more vivid colors. This is essentially a simple 'Levels' adjustment. It's like using a metal stamp 8 inches wide and cutting the edges off a 12 inch wide ruler.
By introducing in-camera settings, the camera squeezes and twists that 12 inch wide ruler a bit so more information gets into the same space. This could be termed as a "Levels" and "Curves" adjustment.
The advantage is that because you have done the adjustments BEFORE taking your 8 bit stamp sample, there is less overlapping and 'gapping' of information.
So in practice, doing major adjustments in RAW then converting to 8 bit JPG can lead to a pretty major advantage over simply shooting and editing 8 bit JPG's, even though the final color space is the same. So my above post should not actually be read that I am agreeing with you that just because you end up converting to 8 bit JPG, you lose the advantage of shooting in RAW. 8 bit JPG is quite capable of producing very acceptable prints.
Originally posted by jhonan:
Originally posted by me:
As you pointed out, if your RAW software converts to an 8-bit image, you've just lost the only advantage to shooting RAW: The ability to edit the image using all 12/14 bits of per-pixel colour depth. |
|
Ultimately, the reason for this is because regardless of how many bits you are dealing with, you are always faced with the same boundary limits of 0 and FULL, or Black and White.
...anyhow... :) |
|
|
|
09/18/2007 08:51:31 AM · #40 |
Gosh david! you opened my eyes
I used to use DPP from time to time, especially for quick B+W conversion: I never realised it was pre-converting to 8 bit
I did a couple of test myself and it seems quite the case. That defeats the whole point about shooting in RAW!!
|
|
|
|
09/18/2007 10:05:37 AM · #41 |
I have been using Canon DPP for some time now and find it to be a capable tool for it's intended purpose. DPP seems to get more out of a RAW image than just about any other converter I have tried, but note that I have not tried Capture One do to the price.
If you want to see the advantage to RAW I recommend that try the following:
1. Do you white balance and basic exposure adjustments in DPP
2. Send the image to photoshop or export and 16-bit tiff and then edit in photoshop.
3. Do a levels or curves adjustement. You will notice that 16-bit image does not get the same jagged edged result that you get with jpeg.
Over exposed is overexposed and black is black. RAW gives you more data in between so there is more latitude within the boundaries of unexposed and blown out.
|
|
|
|
09/18/2007 10:49:04 PM · #42 |
mouten. Read closer.
This appears to be the case only for SOME RAW conversion programs. shooting in RAW is still useful, both for doing major edits as detailed below and for rescuing highlights and shadows in software that is capable of this - Adobe Products appear to fit the bill here.
Nusbaum. you are talking about the advantages of editing in 16 bit space rather than 8 bit space, not the advantage of Canon DPP software itself. The fact is that over-exposed is not necessarily hopeless because a little rescuing CAN recover some lost detail (when compared to the equivalent JPG) at the edges or the histogram, but your converter has to be able to see the entire RAW file. This is a claim by the software manufacturers as well as many people who shoot in RAW. See examples above, as well as the histogram while working with Adobe Camera Raw yourself.
I too have been using DPP for a while, although I haven't done a lot of work with trying to recover stuff that appears blown. 5 minutes working with Adobe Camera Raw really opened my eyes as to what is possible. |
|
|
|
09/18/2007 11:42:09 PM · #43 |
changed my mind
Message edited by author 2007-09-18 23:44:25.
|
|
|
|
09/19/2007 04:27:11 AM · #44 |
Originally posted by eschelar:
mouten. Read closer.
This appears to be the case only for SOME RAW conversion programs. shooting in RAW is still useful, both for doing major edits as detailed below and for rescuing highlights and shadows in software that is capable of this - Adobe Products appear to fit the bill here. |
sorry I was not clear. I use ACR most of the time. And it does a great job at recover burnt highlights etc. in raw. It is also in many other ways more powerful than DPP. But I would still use DPP from time to time for quick B+W conversion, especially when I shot with the B+W setting in camera.
Now I will only use ACR! |
|
|
|
09/19/2007 04:53:17 AM · #45 |
In my opinion, shooting RAW for converting to B&W is useful regardless of whether you have the extra ability of recovering blown highlights or not because I find that the extra internal information provides smoother transitions, particularly when really pushing specific channels to really make it punchier. This follows some of the same reasoning why the greens channel is often a cleaner channel than the others for converting to B&W (50% of the sensor photosites are green, compared to 25% each for Blue and Red, so there is more real information there).
I believe that ACR is capable of detecting camera settings including B&W. I would imagine that this would include the 'color filter' B&W effects in some more recent camera models (my 30D has these). |
|
|
|
09/27/2007 09:29:23 PM · #46 |
Just a quick update on this thread. The question is still with Canon support. In case anyone is still interested in this thread, here's another webpage I made for Canon, with another test image...
//www.surfplan.com.au/temp/DPPRAW.html
It shows the problem better, and also shows that DPP is not completely useless, because it does give a nice sky gradient that couldn't be obtained from the JPG, although there is still no extra highlight information. Interestingly, adjusting the white balance was able to extend the gradient just a little further to almost touch the trees, but this extra highlight data should be accessible through brightness adjustments as well, not just with white balance adjustments.
As another side note, I have used DPP a little more with some other test images. I can see benefit to using it for adjustment, and the resulting images are SO sharp. Problem is, the workflow gets a lot more complicated when dealing with adjustments that modify the original file, especially when DPC validation is concerned. Hey, did somebody say Lightroom? :)
Anyway, I'll post the results on DPP when I hear back from Canon.
|
|
|
|
09/27/2007 09:48:42 PM · #47 |
Incidentally, in my understanding, DPP does not adjust the original file. It tags a bunch of adjustments on to the file in it's own cache. Similar to the .xmp tag files that Adobe uses. Even white balance has a tag. this is possible because it is a RAW file (not an image file). Shouldn't be a problem with validation.
You can see this if you need to revert to the original image or reset a particular setting to shot originals.
If I recall correctly, shots that are modified have a star in the DPP viewer. The star indicates that the DPP viewer has stored additional information for it. Load that same RAW file into adobe and this information is ignored because it is in a different format. |
|
|
|
09/27/2007 09:57:13 PM · #48 |
DPP does non-destructive editing in the sense that any edits do not change the pixel data in the RAW file, and are just extra tags. The original shooting tags are also retained, so you can revert at any time to "as shot" settigns.
However, when you close DPP, it saves the tags into the RAW file. You can see this by the date/time stamp of the RAW file, which changes when you exit DPP and save the changes. So the original file is modified, and by my understanding, this makes it no longer acceptable for DPC validation. They won't accept a JPG file that has modified tags, so I don't think they'll accept a RAW file that has been modified in this way either.
|
|
|
|
09/27/2007 10:18:58 PM · #49 |
| Interesting. I may stand corrected! zipes! |
|
