HDR and large format?

[Tyler Boley]

I think here and in the Perceptool thread people are just trying to show how a process helped (helped, not solved) with a difficult technical problem, generously sharing something they learned.
So whether or not everyone connects with examples artistically is irrelevant. It's questionable whether or not examples have been shown are even implied to be successful finals, only that particular issues were addressed.
Critiques irrelevant to the thread just discourages members from interacting, which is the real point of these lists. I for one appreciate every suggestion and example offered here.
Tyler
http://www.custom-digital.com/

[John Brady]

John, if you are not already familiar with Really Right Stuff products, their pano-mount might be the ticket.

http://reallyrightstuff.com/pano/07.html

Also, this, or a similar product, might be useful when doing two-shot panoramas with your new TS lens. Start with the lens shifted all the way left, then shift right and adjust the body so that the nodal point is where it was for the first image.

http://reallyrightstuff.com/macro/02.html

I agree they have some great stuff. I use their qr clamps and plates. I got one of their L brackets for the 5d.

My plan was to do as you have described Kirk. So far I have played with setting the camera up vertically (portrait) and firing three shots, left shift, center and right shift. The field of view with this 17mm lens and full frame is incredible. I tried just 2 images left and right but there was some missing area in the middle.

Now the next trick will be to do the above but also make multiple exposures and try some HDR.

jb

[Wallace_Billingham]

No, you are missing the point, all my files are shot at ISO 200. It has nothing to do with high ISO. I don't bracket with ISO ever. The low noise of exposure blended files is because of the healthy exposure for the shadows. The bracketed exposure for the shadows has virtually no noise in the shadows because the shadows are placed high like on Zone V, and that is the file that the HDR program uses for the shadows. Whatever noise you have in the boosted shadow exposure is what you will get in the file of the blended file.

Try it. It is obvious.
__________________

but that is not the way HDR software works. HDR was designed for use in computer generated animation for both videos you watch on a screen and video games. The idea was to have a very high bit depth file that could contain a lot of information at the pixel level about how that pixel could be rendered in a variety of simulated lighting effects. This data was then tone mapped on the fly so that if the animation called for say a robot in bright sunlight it would map the tones much brighter than the same robot would be in the dark. You would not have to have dozens of different models of the robot in various lighting conditions you would just have one model that would be rendered differently.

When used in still photography each exposure gets put into the data at the pixel level. Nothing gets tossed out every bit of data goes into the file which is why you will get better results if you have say 5 exposures than 3. The HDR software looks at the images and sees how the scene would look in variety or lighting conditions from very bright light (over exposed) to very dim light (under exposed). All of that data gets loaded into the program noise and all. That is why the software will do a simple noise reduction before hand by reading from the exif data what the ISO was. It also works better when it can read from the exif data what the shutter speeds were so it can better know what file is what.

It does not toss out the under and over exposed data with higher noise, it all goes into the file so that when you render the image it is able to have as much data as possible. That is why it is very easy to get very fake looking cartoonish images with HDR software because all of that data is in there, and the vastly over and underexposed images create color shifts

If all it was doing was using the properly exposed pixels you would not get the color shifts just like you do not get a color shift if you take two exposures and blend them together using a mask in photoshop.

You are getting very low noise files because you are starting with very low noise files and you have increased bit depth to render shadows without artifacts. The program is able to render the shawdows with a lot more data. If you are doing a 3 shots a stop apart in HDR the shadows that would normally fall in zone 2 also get data from those shadows being rendered in zone 3 and 4. All of that data then slides back down into zone 2 if that is how you choose to tone map the image.

When you are dealing with film scans however you have the same exposures in the digital file created with the imagers in the scanner. You will end up properly exposed digital files of the negatives/positives placed in front of the scanners imagers. The scanner and the HDR software will have no way of knowing what was over exposed and what was not. Is has no need to as all of the data gets loaded into file and then gets rendered. If shot one had a pixel that recorded at zone 1 and shot two had the same pixel that got recorded at zone 2 and shot three had the same pixel recorded at zone 3 you as the person rendering the file would have to tell the program (or let it guess if you use one of the automated renderings) what zone to put that pixel on in the final image. Of course the program would also have no way of knowing if that was a pixel of pure data or a pixel of noise. How would it?

[George Stewart]

kirk and Emmanuel, I concur and I do not concur. The "entrance pupil" is the correct term for the point at which a lens' axis must be rotated, or remain stationary, in order to avoid parallax . However, with respect to the TS lenses (Nikon 24, 45, & 85 PCs) I have used, the entrance pupil does move when the lens is shifted, hence requiring a corresponding but opposite shift in the camera body. While the effect may be negligible depending on subject distance, it is clearly visible through the viewfinder on close subjects.

John, if just shifting the lens works for you, great, otherwise consider also shifting the camera body. Finally, how is shooting in southern Florida during the summer (mosquitos)? Does Corkscrew have water? I'll be down there in about two weeks.

[Kirk Gittings]

I'm confused. It looks to me like you are saying the same thing basically I am....sort of.

So without the high bracket for the shadow detail (which has lower noise in the shadow areas of the normal exposure because of the increased exposure right?) what noise would you get in the shadows of the HDR? Or to put it another way-rather than using the high bracket (with the increased exposure in the shadow areas) and instead using the middle file and simply boosting its exposure in ACR to artificially create that bracket (which will increase shadow noise). Which method would give you less noise in the shadows of the HDR? The one with the proper high exposure bracket right? So how can you say that the high bracket with the less shadow noise does not account for the less shadow noise in a properly bracketed HDR. Help me out here.

[John Brady]

John, if just shifting the lens works for you, great, otherwise consider also shifting the camera body. Finally, how is shooting in southern Florida during the summer (mosquitos)? Does Corkscrew have water? I'll be down there in about two weeks.

George, Summer is my favorite time to photograph here. This is the time we get the most dramatic skies, best sunsets, best light, etc.

But the mosquitoes can be a pain. We have these little gems called no-see-ums that are a real treat when out on the islands.

Corkscrew is about 7 miles from my house, I haven't been there this year but based on the conditions of most of the surrounding swamps it should have water. It was bone dry for a long time. I don't go there very often because they don't let you get off the board walk. If you're feeling more adventurous than that give me a shout and I will give you some of my hotspots. I have a small gallery here you can check out if you get bored too.
www.gladesgallery.com

[Wallace_Billingham]

I'm confused. It looks to me like you are saying the same thing basically I am....sort of.

So without the high bracket for the shadow detail (which has lower noise in the shadow areas of the normal exposure because of the increased exposure right?) what noise would you get in the shadows of the HDR? Or to put it another way-rather than using the high bracket (with the increased exposure in the shadow areas) and instead using the middle file and simply boosting its exposure in ACR to artificially create that bracket (which will increase shadow noise). Which method would give you less noise in the shadows of the HDR? The one with the proper high exposure bracket right? So how can you say that the high bracket with the less shadow noise does not account for the less shadow noise in a properly bracketed HDR. Help me out here.

Kirk,

You and I are in the same book and chapter but not on the same page. You are refering to a digital file from start to finish and I an refering to the original poster where he is going to use film scans.

When you use a digital file from a DSLR or any other digital camera there is exif data the HDR program uses that exif data to build a very high bit depth model of what the photograph looks like under various lighting conditions. The program can read the exif data and makes a guess as to which ones were over exposed and which were under exposed. However all of the pixels from each file gets entered into the model. So yes noise from under/over exposed files gets entered into the programs model as well. The difference however unlike when you adjust the exposure from a single RAW file is that the noise in the shadows stays low because it gets burried in darker tones and colors and it does not get amplified by trying to slide it to the right of the histogram, because all of the data gets entered into the model the program will apply noise reduction based on the ISO of the exif file before those files get entered into the model.

Now when you render the output file or image from the HDR model the program creates all new pixels based on what output paramaters you have selected. It does not recycle or use any of the pixels from the original files but creates all new ones. This is why it is so easy to very funky and strange colors and other odd looking images because your computer is making an entirely new image right down to the pixel levels.

The problem with using film scans is that you have no exif data, and you do not have over exposed and under exposed files. What you have is a 2nd generation image that is exposed properly of a negative or tranny that was not exposed properly. So if you have blown out highlights the data at the pixel level in the scan will just render that as white. There will be no details just white just like on a darkroom print there would just be the white of the paper. With shadows you will have a very thin tranny or negative one that the scanner will try and pull data off of and thus introduce noise at the scanner itself. The scanner operator could slide the black point over to render all of that as black but that will make it harder for the HDR program to render the tonemap of the image and may also induce noise. It will not be a lot of noise but it is still there.

[cjbroadbent]

....So are any of you employing this method or do you know of any well done examples out there? Please share some images if you have any...

Sorry if you've seen this before. I tried HDR tone-mapping on seperate 4x5 negatives and on a two-scan single 8x10 negative. The two-stop bracketed 4x5s work better than anything I can do in the darkroom. The single 8x10 has nothing much to show for the effort. The tests are on this page:
http://picasaweb.google.com/cjbroadb...eat=directlink
Disregard the digital camera tone-maps which are there for comparison (and which work much better).