Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

HDR imaging

Trevor Meier

Trevor Meier

Well-known member
Joined
Jan 2, 2007
Messages
855
Reaction score
1
Points
0
Website
www.trevormeier.com
I've thought of another reason DNG support in RedCine would be useful...

http://www.flickr.com/photo_zoom.gne?id=387019269&size=l

(This guy's stuff has been popping up a lot lately.)

The ability to convert RedCode RAW files to DNG would allow me to process them in an HDR imaging tool like what's used to create these shots. Yes, I realize HDR 'live action' images are typically multiple exposures, but a similar effect can be created with a single RAW image (For example with Photomatix)

This is a highly stylized look... only useful in certain applications, but the only practical method to realize it without a trick in the camera is by utilizing the dynamic range available in a RAW image file.... and it doesn't make sense to create this type of renderer inside RedCine.

So, can one person get two votes for DNG in RedCine?
 
i agree that HDR has it uses but not to that extent, there's something wrong in the aesthetics of that image (maybe it's just me).

luis
 
For these kinds of images you would not need any raw output? Just output a 16-bit (RGB) TIFF and load that (the app you mention supports loading TIFF files)
 
luis bustamante said:
i agree that HDR has it uses but not to that extent, there's something wong in the aesthetics of that image (maybe it's just me).

luis
Click to expand...

Apart from no proper shadows in the gorge...and all the other 3d graphic stuff that makes it look Soooooo sci-fi, i'd love to play with HDR.

If your shooting say two 1/100 sec 'back to back' frames as a HDR 'frame pair' isn't that effectively one 1/50 combined frame, with a whole lot of sensor maths/ weird motion blurring going on. Is combining those framed pairs a mathematical nightmare?

DF
 
Of course DNG conversion would be faster and significantly smaller than 16-bit TIFFs, technically there are a good way to go (even the SI-2K has a DNG option.) The only potential disadvantage I can see would be marketing, if you have a lot of signal processing as part of the demosaic to 16-bit TIFFs, the DNG would bypass that allowing the pluses and minuses of REDCODE RAW and the sensor data to be directly "exposed". I can understand wanting to make you output as pretty as possible, although even a filtered DNG output would have speed and size advantages.
 
AFAIK that software requires multiple TIFF exposures, or can work from a single RAW file. Please correct me if I'm wrong...
 
little knowledge is a dangerous thing.....

little knowledge is a dangerous thing.....

Saw this paper which mentioned a few (6) tuning technigue to increase DR. Be buggered if I'm gonna follow (remember)the math or squeeze RED IP outta Graeme on this one, but nice article all the same.

http://www-isl.stanford.edu/~abbas/group/papers_and_pub/isscc02_tutorial.pdf

Still, love the idea of getting 15 stops outta RED without it looking like some Pixar movie.

DF
 
Rob Lohman said:
For these kinds of images you would not need any raw output? Just output a 16-bit (RGB) TIFF and load that (the app you mention supports loading TIFF files)
Click to expand...

Hi Rob,
I've start some times ago a thread with the same subject. (here: http://www.reduser.net/forum/showthread.php?t=528 )

Just because the guys of our "Photo" department start to ask me (!) this feature...

I think the main reason is merely the workflow involved. Open the source Raw images with Photoshop (for example) will be a dream (at least for us).

There is a small space for an hope?

Thanks for all...

Corrado.
 
You could also use OpenEXR files (that contain the huge dynamic range of a RAW file) and the simple command-prompt tools that are avalable for unix users (os x on intel is, after all, a unix system).. does REDCODE support Open EXR? I ask because it's ILM's file format of choice (after all, they did make it)

command line tools:
http://scanline.ca/exrtools/

You'd still need to CC it a bit, but it would simplify the tone mapping process by using already-avalable algorythms, saving you from having to do it manually.

Gavin, you mentioned doing it in Shake, know of any implementation (i.e. readymade nodes/scripts), or are we on our own integrating it?

-Wade
 
HDR in the hands of most people is abused and is generally considered to be a joke by most serious photogs. Although, of course, there are exceptions to this. Generally, whenever I see HDR on Flickr I run in the other direction as fast as I can.

Also, moving HDR images are very strange, and tend to flicker and wig out a lot, because the computer, I suppose, does not merge the photos exactly the same as they change from frame to frame.

So basically you are saying that any RAW images can be HDR'd? The software or whatever just generates three different JPGs from the RAW, then merges them? That's kind of interesting. All my photos are RAW, so maybe I will try it out. Are there any free programs that do this? Or Photoshop?
 
Well if you want the "HDR look" all you have to do is just adjust the curves so that the luminosity is relatively constant. Assuming of course the data hasn't clipped.

If you have multiple exposures and you still want to make a go of it in Shake, just apply a luma key to both files and play with the curves... it's not exactly a complicated tree.

There is no reason to Break up a file, and then remerge it. You'll only lose data. The "HDR" "Tonemapped" photos that you see online are just running a proprietary algorithm that screws with the luminance based on relative brightness (i.e large bright patches get burned down to neutral brightness, and dark patches get dodge up to a neutral brightness). You can do the same thing through curves + dodge/burn manually.

The only unique aspect of some HDR file formats is that they actually encode a realworld luminance instead of a relative RGB value for example a .rad file. Other than that they're just high bit depth RGB images. However even that's not true technically you could have a 4bit HDR image. It just would be pretty much worthless.
 
Personally, I'm less interested in the cheese-fest Flickr-type stuff, but more interested in having a contrast effect useful for a particular creative look.

Gavin -- I meant an integration of some sort of algorythm to make it more "scientific" (or at least elegant). I'm going to try an implementation of this one... you can see example process of what i'm looking for by referencing the end of the paper against the EXR linked after it to see what I'm after -- merely a "realistic" (as opposed to oversaturated flickr crap with halos) range compression for difficult images (due to bad lighting, etc)

The paper with the algorythm (simple luminance avg based on relative brightness, as Gavin said):
http://graphics.cs.ucf.edu/papers/cic2005.pdf

Image of church referenced in paper (in unprocessed EXR)
http://scanline.ca/exrtools/memorial.exr

Anyway, to place this in context of the topic, I think that this kind of thing is simple to do with images that REDCODE will put out anyway (probably DPX) via Shake or whatnot, and yes, you could use Automator and Photoshop to do the exact same thing if you wanted... DNG output isn't really needed, just some sort of high-dynamic range file format, and I think DPX already satisfies that (not too sure -- any of you more techy than me feel free to contradict).

-Wade

Edited for bad grammar.

-W
 
Check this thread for my results with HDR-ish (but not HDR) tone mapping. Thanks for the idea Gavin!
 
You must log in or register to reply here.
Back
Top