More info...
- Thread starter Jannard
- Start date
Jaime Valles
Veteran
Nice. The detail in the milk bottle is impressive.
A digital negative, indeed.
A digital negative, indeed.
Muttondraw
Well-known member
The histograms look loads better and I think the slightly less blue colour balance is more in keeping with the subject too. Nice one Jim, only wish I could make it to the screening to see this in motion.
Martin
Martin
AuditoryVisuals
Well-known member
It still looks natrual and the highlights are under control. So I guess the RED ONE captures a wide dynamic range picture, but you have to moderate it?
It looks a lot better than the original.
It looks a lot better than the original.
laboprod
Active member
Elo Jim,
Don't be sorry man, nobody here will argue on the luck we have to share so much of the dev work with the RED team....
Indeed, this image is better, and I think you have changed the 2k version on your web site as well.
I couldn't find a pixel which is "burn out", which was the case on the previous image on several spots (collar, milk bottle, sandwich paper). As you say, you preserve more data in the highlights. I have to say that, as Jaime, I think that the details in the milk bottle are impressive and reflect a good diynamic range.
Thanks (for the thousand times
antoine.
Don't be sorry man, nobody here will argue on the luck we have to share so much of the dev work with the RED team....
Indeed, this image is better, and I think you have changed the 2k version on your web site as well.
I couldn't find a pixel which is "burn out", which was the case on the previous image on several spots (collar, milk bottle, sandwich paper). As you say, you preserve more data in the highlights. I have to say that, as Jaime, I think that the details in the milk bottle are impressive and reflect a good diynamic range.
Thanks (for the thousand times
antoine.
Last edited:
Metaphorpix
Member
Looks great on the second pass!
-Dave Dessel
-Dave Dessel
Mr. Blonde
RED Advocate
That shot's money. :thumbup:
imgentertainment@mac
The Reverend
The leavels look really good. Has allot of range. and Yes the milk is there and not just a white blob good work guys
Graeme_Nattress
Red Team
Perversely, one of the big problems with a wide dynamic range, is how do you map it into a viewable range. That's the problem we had with this image, and then Jim looked at it again, and got a much better image. That's the benefit of raw that those decisions are not baked into the image.
Graeme
Graeme
glenn chan
Well-known member
What I think film does:
Each channel has a s-shaped transfer curve/function. You can emulate this in photoshop by using adjustments --> curves. make the curve s-shaped.
If you look at Kodak's data sheets for their films, you get an idea of what this curve looks like.
However, each layer in film interacts with each other. High density in one layer will reduce the light passing through itself. So, transmission is cumulative.
For each wavelength:
Final light = Input * layer1's transmission % * layer2's transmission % etc...
The actual math gets complicated even if you split all visible wavelengths into 25 bands (or some other number around there). You have to do calculations for each band of wavelengths, and then figure out the final perceived colors based on the CIE CMFs. Four layer films (i.e. Fuji) also add complication.
Film's transfer response could be simplified into a 3-D look-up table. I think you'd have to assume 3 wavelengths of light.
2- One thing you may not like about film's transfer characteristic is that it shifts hue and saturation, especially in the highlights. It does the same in the shadows, but it's less visible. You can sort of play around with this in Photoshop.
Some modern film stocks have much longer linear segments, so that area sort of responds more like video.
3- The following PDF discusses some other methods of mapping extended/high dynamic range images into viewable range (this is from the academic side, but shows what's possible):
http://web.mit.edu/persci/people/adelson/documents/Li_ACM_24-3.pdf
Each channel has a s-shaped transfer curve/function. You can emulate this in photoshop by using adjustments --> curves. make the curve s-shaped.
If you look at Kodak's data sheets for their films, you get an idea of what this curve looks like.
However, each layer in film interacts with each other. High density in one layer will reduce the light passing through itself. So, transmission is cumulative.
For each wavelength:
Final light = Input * layer1's transmission % * layer2's transmission % etc...
The actual math gets complicated even if you split all visible wavelengths into 25 bands (or some other number around there). You have to do calculations for each band of wavelengths, and then figure out the final perceived colors based on the CIE CMFs. Four layer films (i.e. Fuji) also add complication.
Film's transfer response could be simplified into a 3-D look-up table. I think you'd have to assume 3 wavelengths of light.
2- One thing you may not like about film's transfer characteristic is that it shifts hue and saturation, especially in the highlights. It does the same in the shadows, but it's less visible. You can sort of play around with this in Photoshop.
Some modern film stocks have much longer linear segments, so that area sort of responds more like video.
3- The following PDF discusses some other methods of mapping extended/high dynamic range images into viewable range (this is from the academic side, but shows what's possible):
http://web.mit.edu/persci/people/adelson/documents/Li_ACM_24-3.pdf
Poi Boy
Well-known member
"How hard can it be if film can do it photochemically? On the other hand jpeg engines of DSLRs can't do a good job of it. Makes no sense to me."
I think dslr do do a good job of it for the most part but are no match for a skilled person with a raw file. I did custom processing and printing for years with still film and while technically film has greater latitude I think you can achieve way more looks with a good raw file.
Aloha
-A
I think dslr do do a good job of it for the most part but are no match for a skilled person with a raw file. I did custom processing and printing for years with still film and while technically film has greater latitude I think you can achieve way more looks with a good raw file.
Aloha
-A
Mardi_Gras
Member
Nice...
Nice...
Nice...
Nice...
Nice...
im.thatoneguy
Well-known member
How about compressing all of the 12bit image into an 8bit? That way we can get a closer look at the raw image.... or release a 16bit tiff on bittorrent.
ProLost
Well-known member
Nicely stated Graeme, and a message that Red will continulally need to help propeate. RAW images should not "look good." They should look low-contrast and underexposed. This is something that DSLR shooters understand but video folks are still warming up to.Graeme_Nattress said:
-Stu
ProLost
Well-known member
Indeed, and in fact the beginnings of such a system is included in Adobe After Effects 7.0. I wrote about it here.glenn chan said:
Just for fun, I used this functionality to remap Jim's image to look as if it had been shot on Kodak 5218 and printed to Vision stock. I describe this technique in my book, btw.
-Stu
I keep thinking these stills look better when sharpened, but like other members have pointed out sharpening could add some very bad effects. I've been told that RedCine would allow the user to decide how much digital sharpening will be done. This brings up a question. Will there be very fine controls in determining the sharpness of the image or will it be more of a sharpen, sharpen more, sharpen even more, type of selection?