Dynamic Rage: RAW vs. JPGs

[rraman]

I have this doubt: does shooting in RAW give more dynamic range than while shooting in JPG? I understand that one may simulate higher dynamic range by making two conversions from RAW, one for highlights and one for shadows and then merging them.

But if this technique is not used, will a RAW file have better DR than a JPG shot with identical settings?

[commando]

I know RAW has 12 bit range whereas JPG has 8 bit. That might make a little difference if you stuff the exposure up a little.

[Hellashot]

RAW has data recorded for 16bit color. Only new/newer printing/editing software supports 16bit color.

Also JPG is compressed data at 10:1 I believe and RAW is only 2:1

[commando]

I only know what i've read. Have a look at this page (http://www.kenrockwell.com/tech/raw.htm). I'm not 100% sure it applies to 10D/300D but I suspect it does.


...

If you're a tweaker you'd be interested to learn RAW and JPG also have the same effective bit precision. JPG has 8 bits per color per pixel and RAW may have 12 bits, but here's the big catch: RAW is 12 bit linear, and JPG is 8 bit log, gamma corrected or some other non-linear transform derived from the 12 bit linear data. Thus in the shadows where this might matter the two are the same, since the full 12 bit resolution in the dark areas is preserved by the non-linear coding. Even if the two formats differed in dark resolution the sensor noise is still greater than one LSB anyway making it a moot point.

RAW is designed for people who intend to spend a lot of time twiddling with one image at a time. For these applications I use large format 4x5" film instead for much better quality, thus you see why I don't use RAW. RAW is very popular for people shooting landscapes with digital cameras, which is not what digital cameras are for. Unfortunately this is becoming popular among amateurs, but remember that for amateurs the fun is in making the photo, and for fun digital is king. Personally I focus on the final image, for which big film excels. This goes to explain all the heat I get from hobbyists, who of course love RAW and shoot Yosemite on their digital SLRs which is very different from what I do. RAW is also for people who don't yet get perfect exposure right when they make the image, although of course JPEGs also allow these adjustments.

...

[Tom W]

I think you'll find plenty of opinion supporting RAW here, and plenty of evidence that contradicts what Ken Rockwell has to say. There's a lot more to salvage from a "less-than-ideally" exposed RAW image than from a JPEG, due to the ability to reach deeper into the shadows and a bit into the highlights as well. In RAW, you can stretch, compress, or move your exposure up or down to a greater extent without hitting the limits of the data.

You basically have extra leeway to pull a great 8-bit JPG out of 12 bits of data.

[commando]

I think you'll find plenty of opinion supporting RAW here, and plenty of evidence that contradicts what Ken Rockwell has to say. There's a lot more to salvage from a "less-than-ideally" exposed RAW image than from a JPEG, due to the ability to reach deeper into the shadows and a bit into the highlights as well. In RAW, you can stretch, compress, or move your exposure up or down to a greater extent without hitting the limits of the data.

You basically have extra leeway to pull a great 8-bit JPG out of 12 bits of data.

I thought that's what the quote from him I copied said? Either way, RAW is good, if you can be bothered working out how to process it.

[FlyingPete]

...RAW is good, if you can be bothered working out how to process it.

Or have the time to do so. If I am out on a job and shoot 1000 shots in a weekend, and have to have them submitted by the Tuesday after, I don't want to add another step into my process!

The single best screw up RAW is good for is having the wrong white balance ser, some JPG's are almost impossible to recover from an incorrect white balance whithout a signifcant loss in 'quality'.

[tofuboy]

I have this doubt: does shooting in RAW give more dynamic range than while shooting in JPG? I understand that one may simulate higher dynamic range by making two conversions from RAW, one for highlights and one for shadows and then merging them.

But if this technique is not used, will a RAW file have better DR than a JPG shot with identical settings?

Here is my guess/input/opinion/(call it whatever you want)...

The dynamic range of jpg vs raw is different... as others have pointed out. You have a lot more room to work with in a RAW vs. a jpg. However, if you don't use that room to or advantage (which is pretty much what your question is asking as I read it), then one could argue that they are the same with that regard.

Shooting in RAW, you can mimic a larger dynamic range than what the digital sensor allows (by using a 2 exposure technique... which really only applies to your typical landscape image with a foreground and a sky that are seperated by a horizon.)

[Tom W]

Rockwell claims "Thus in the shadows where this might matter the two are the same, since the full 12 bit resolution in the dark areas is preserved by the non-linear coding. Even if the two formats differed in dark resolution the sensor noise is still greater than one LSB anyway making it a moot point."

I don't agree with him in this respect - there's simply more data to work with in RAW, and his contention that the linearity of one vs. the logrithmic nature of the other somehow makes them the same doesn't prove true in practice.

You CAN pull data out of shadows in RAW that is lost in JPG.

I was also a little less than happy that he has proclaimed that digital cameras are not made for shooting landscapes. They're made to shoot anything with visible light.

Anyway, to make a long story short, I agree with you, but I don't think that Ken Rockwell does. At least not fully.

[commando]

Anyway, to make a long story short, I agree with you, but I don't think that Ken Rockwell does. At least not fully.

I didn't really read everything he said, i'm a bit busy today to pay attention.

[Steven M. Anthony]

There is a great book on the subject--"Camera Raw" by Bruce Fraser. The jpeg output from, say, a 10D STARTS as a raw file--in the sense that the 10D's sensor records the number of photos striking it. Then, IN THE CAMERA, there are a bunch of processing steps that occur (some of which you can direct through "paramaters") and some compression (that you can also direct, at a gross level, via the Quality function (e.g., large/fine, large, medium fine, etc.)).

Once these processing steps are taken, they cannot be reversed.

You have much more control over all the processing of a raw file.

One interesting note for the book that is relevant here: When you shoot in RAW format, the image you see on the LCD is generated from the companion jpeg that is created with the raw file. If you use the info screen, you might see blinking areas--indicating blown out highlights (in the jpeg). Fraser points out that when you look at your raw file, you will see LESS blown out area, because the file is not forced through the compression the jpeg received in-camera.

[rraman]

Thanks a lot guys for all the informative posts!

[nemesis099]

There is a great book on the subject--"Camera Raw" by Bruce Fraser.

is this the book you are talking about Link (http://search.barnesandnoble.com/booksearch/isbnInquiry.asp?userid=Mo5dgZo9VH&isbn=032127878X& itm=1)?

I will probably take a look at it when I get the chance.


As far as my opinion with RAW and JPG. I would like it where the camera takes a JPG with the settings and keeps the RAW file also. I believe most cameras do this so you can view them on the LCD but the JPGs are hard to get to.

This would allow you to look at the JPG and if that is the picture you wanted you are finished with it. If you aren't happy with it you can look at the RAW file and see if you can touch it up for use.

If I am not mistaken the new Canon 20D does store these files individually rather then under 1 file. (Correct me if I am wrong)

Just my opinion because I know that when I take a ton of pictures some come out close enough to the way I wanted and then some just aren't right and I wish I had the RAW to play with. Of course then when I take all RAW pictures I can't stand the fact that I have to use a program to develop them.

just my .02 cents


edit: fix link

[Steven M. Anthony]

n099: Yes, that's the book. My bet is that if you get the book, you will become convinced to always shoot RAW and process through Camera Raw. It is really easy and the results are awesome (at least compared to what I had used before). What really helps vs. the canon File Viewer Utility is color histogram and "real-time" changes with the slider controls--no waiting for the Utility to re-generate the image. Camera Raw also has features that show you--on the image itself--which highlights and/or shadows you are clipping while you adjust exposer settings.

As the author points out, once the Raw file is converted to jpeg or tiff, any post-processing will, at some level, degrade the quality of the image. No one processing step might compromise your image, but the additive effects of several processing steps might. But the adjustment to the Raw file done in Camera Raw--according to the author--either do not degrade the image at all, or, in the case of major shifts in shadow exposers changes, will degrade the image far less than Photoshop post-processing.

The book is well worth the price if you are new to working with Raw files. I just got the book on Saturday, and downloaded the trial version of PS CS on Sunday. I've worked with a couple of images I had worked on through PaintShop Pro V9.0 (but starting with the Raw file again). Most of what I did to the images in PSP9 I could do in Camera Raw (except PSP9 doesn't work with 16bit images). When I sharpened in PSCS, I was able to use threshold = 0, radius = 0.5, and amount = 500% with no noticable artifacts when printed at 11 X 17 on an Epson 2200 (on Premium Luster). If what I had done on PSP9 was a "9 out of 10," then what I did through Camera Raw + PSCS was a "15!"

[meow]

A picture says more than a thousand words, they say. :lol:

As I understand it RAW doesn't extend the dynamic range, strictly speaking. That would depend on the camera. It does give you more levels/shades/tones (whatever it should be called) to work with.

Now I'm going to prove that RAW, or at least 48 bit, owns. :mrgreen:

See here this ugly little crop from an image converted to 48 bit TIFF from RAW.

http://ababa.net/stuff/bitdepth/48crop.jpg

I converted that image to 24 bit. Then I converted THAT image back to 48 bit. I now had 3 images. The original 48 bit crop, a 24 bit image and a 48-to-24-to-48 bit image.


I applied this curve to all three images.

http://ababa.net/stuff/bitdepth/applied_curve.jpg


Resulting in three little images looking something like so:

http://ababa.net/stuff/bitdepth/48crop_after.jpg



Now lets check the curves of these three images.

The original 48 bit one:

http://ababa.net/stuff/bitdepth/curve_48.jpg

The 48-to-24 bit one:

http://ababa.net/stuff/bitdepth/curve_48-24.jpg

The 48-to-24-to-48 bit one:

http://ababa.net/stuff/bitdepth/curve_48-24-48.jpg



Not exactly brain surgery, but it's good enough for me. I rest my case. 8) :lol: