HDR question

JR
Posted By
jouko_ruuskanen
May 4, 2005
Views
737
Replies
24
Status
Closed
I’m trying to merge several ACR conversions (different exposure compensation values) using "Merge to HRD" function, and I always get "There’s not enough dynamic range to perform the conversion"…. The images I use are 16 bit tiff files, and they blend very well using something like PhotoMatix… not PS. Does HDR somehow check the EXIF data, and conclude there’s no sense doing the conversion? If yes, I consider this as a serious flaw in the function….
Jouko

MacBook Pro 16” Mockups 🔥

– in 4 materials (clay versions included)

– 12 scenes

– 48 MacBook Pro 16″ mockups

– 6000 x 4500 px

JS
Jeff_Schewe
May 4, 2005
You can’t. You’ll need actual different exposures, not just different processed files from the same exposure. . .you can merge different processed images as layers in 8 or 16 bit, but HDR needs FAR more dynamic range that what changing the raw settings can provide.
JR
jouko_ruuskanen
May 4, 2005
Jeff,
I just tried with two JPEGs that have +1 ev difference, and HDR worked fine. With RAW, you can get +2 ev difference, and HDR refuses to do the job because of "lack of dynamic range". This is ridiculous, HDR should just accept my input and do the work, without trying to stop me in case EXIF happens to be the same. I saved my RAW conversions as JPEGs without EXIF, and merged them using HRD… worked fine, I just had to manually input exposure values! This means I can use HDR with RAW conversions only if I tweak the EXIF values…
Anyway, the free Photomatix Basic seems to do at least as good a job, and it doesn’t complain about EXIFs… if only they had a PS plugin

Jouko
WD
WEA_Davies
May 4, 2005
I have successfully merged, in CS2, three Canon CR2 files that differ overall by 2 EV. However when I try to accept the merger I get a big red cross and the statement that a command is missing. What command is this and why is it missing?

William
RG
Rags Gardner
May 4, 2005
From what I was taught last week, you need 3 to 8 images and they must each be exactly one stop apart as far as exposure. I only tried it with the class lab images.

Cheers, Rags 🙂
Exposure is all about Zone V
RB
Robert_Barnett
May 5, 2005
Chris Cox said you can not do HDR with the same image that has the exposure adjusted a stop or so. The problem is that it is the exact same image with the exact same data. The only thing you did when you adjusted the exposure was make the data easier to see in the case of exposing for the shadows or harder to see in the case of exposing for the highlights. You need to take three shots or more with your camera using different exposures for HDR to work.

Robert
RG
Rags Gardner
May 5, 2005
OK, Julieanne Kost had it wrong or I misunderstood her. My bad.
MD
Michael_D_Sullivan
May 5, 2005
I was successful in using HDR to process three TIFFs produced from the same negative in my scanner. In the scanner software I used curves to produce negatives that included (a) only highlights (top 1/3 of input values), (b) only midtones (middle 1/3), and (c) only shadows (bottom 1/3). In each case, I used a steep straight-line curve to map the chosen range across the entire output range, blowing out any higher pixels and losing any lower pixels. I combined these in HDR, making up exposure values 4 EV apart. It took a bit of processing and tweaking, but it worked. I suspect you could do the same thing in ACR.

The subject, in my case, was a shot of the interior of a broken-down building, with an open roof and snow on the floor, but the majority of the interior being in deep shadow. I had little luck getting the entire usable dynamic range into a single scanned TIFF even with the use of curves, so I tried separating it into three files. My intention was to combine them in CS, using layer masks and curves. That didn’t work too well, but it worked very well using HDR.
DM
dave_milbut
May 5, 2005
that’s an interesting little story… 🙂

thanks for the tip mike.
MD
Michael_D_Sullivan
May 5, 2005
DM
dave_milbut
May 5, 2005
looks good… can you post the 3 originals?
MD
Michael_D_Sullivan
May 5, 2005
Sure. You can view them here <http://avogadro.smugmug.com/gallery/503568/1/21287549>. I also posted a straight scan from the negative to show what I was trying to improve on.

These are all greatly reduced in size, needless to say. I was using 22MB 16-bit TIFFs. I also did a lot of work on the final product after HDR. The negative was not in great condition.
CC
Chris_Cox
May 5, 2005
Yes, MergeToHDR reads EXIF information for exposure values.

No, trying to create an HDR image from a single RAW file is pointless — all the data already fits in 16 bits!
JJ
John Joslin
May 5, 2005
Good tip Mike; maybe I can rescue some less-than-optimum transparencies that way. 🙂
JR
jouko_ruuskanen
May 5, 2005
Chris, all the data in RAW does NOT always fit into 16 bit, at least Canon claims +2 ev latitude. I’m just trying to make the best out of a single RAW.
Jouko
CC
Chris_Cox
May 6, 2005
Jouko – that’s odd, because the RAW data starts out at 12 bits per pixel, so it MUST fit within 16 bits.
SP
Suan_Pin_Foo
May 15, 2005
Just to see the results, I converted one RAW file to three JPEGs with different adjustments in ACR and ran the Merge to HDR. The results were surprisingly good despite the losses in the conversion to JPEG. And this was just a very rough test.

It seems like all the building blocks are there to allow us to specify the exposure manually. Why not have a dialog in the Merge to HDR to ask us if we want to use the EXIF data or we want to specify this ourselves? It gives us more options and control and many of us know what we are doing.

It also benefits Adobe in that we may be able to do things you had not thought was useful or desireable.
G
GordonGraham
May 15, 2005
It seems like there’s a lot of confusion about how this feature actually works. Many folks are familiar with the process of doing two different conversions in ACR at different Exposure values and then combining them in PS using a mask. HDR sounds like it does something like this and so people try and use it that way, without success. Then the explaination comes that it doesn’t quite work that way and you need several different exposures.

Would it help to get a good description from someone that really understands HDR as to exactly how it works and why it won’t work with multiple conversions from a single RAW file.
MD
Michael_D_Sullivan
May 15, 2005
It seems to me that by definition it won’t work for its intended purpose (creating a realistic composite at 32 bit depth based on widely spaced exposures) if the exposures fed to it come from a single RAW file. On the other hand, it could accomplish a useful artistic purpose by using the same mechanism to combine different images from a single RAW file, particularly if those images have been highly tweaked using curves, levels, etc., if one is permitted to override the EXIF metadata. I have done something similar to this, by combining highly tweaked images scanned from a single negative; there was no exposure metadata in the files, so HDR asked what the exposures were and I lied that they were 4 EV apart. I was able to get a useful image out of HDR that was better than what I had been able to generate manually. The same kind of manipulation would be useful for RAW images, even though it won’t result in a true 32-bit image as it was designed to do.
SP
Suan_Pin_Foo
May 15, 2005
I fully agree with Michael that this is "by definition" going beyond intended purposes. Then again, when has anyone of us stopped using a feature just because it was not the intention of the developer?

I can do the same thing a couple of different ways (all of them tedious and not very satisfying). It just seems to me that it would be nice to have the ability to "cheat" CS2 a bit and end up with a HDR file that I can re-use for other stuff.

By the way, the quickie HDR turned out pretty nice despite going to JPEG. I think I may try the same thing with TIFF and check out the results.
RB
Robert_Barnett
May 15, 2005
The problem with the single image is that no matter what you do to it with Levels and Curves or anything else in Photoshop they are all the same image with the exact same data. Levels and Curves only allow the human user to see the data better, Photoshop on the other hand has always seen the data even if the human couldn’t.

This is why HDR needs images takes using difference exposures at the time the pictures were taken. Auto bracketing if your camera has it is one way to do it. This will give you three completely different pictures for HDR to work with. Though for the best results you should…

1. Take a reading in full auto with your camera and make not of the f/stop and shutter speed.
2. Then take a pictures in manual mode setting the shutter speed to full auto amount and then adjusting the f/stop up to max taking a picture at every full stop and then back down (with the full auto amount as the starting point) to the minimum taking a picture at every full stop. This will give you a series of pictures ranging from your cameras minimum f/stop through to its maximum f/stop. Usually around 6 to 8 shots. These need to be taken using a tripod and a remote shutter release so that all 6 to 8 shots can be overlapped perfectly. Even with a good tripod pressing the shutter release manually will cause very unwanted camera movement hence the remote shutter release.

I have managed to get some very nice images with the above technique. My only problem is that after combining to 32-bit they are can of dingy. I have to covert them to 16-bit and use levels and curves on it to get any decent color and contrast. The "Exposure" tool that works in 32-bit mode just pain sucks for this type of thing.

Robert
CC
Chris_Cox
May 16, 2005
Robert – the exposure adjustment has nothing to do with converting to 16 bit.

Are you confusing Exposure & Gamma (one of the 4 toning options in the 16/8 bit conversion dialog) with the Exposure adjustment?
RB
Robert_Barnett
May 17, 2005
No my complaint is that levels and curves isn’t available and that the only option in 32-bit depth is the "exposure" options which I personally think sucks. Now I understand that 32-bit is a pretty new and drastic feature for Adobe and I am willing to cut some slack. But please for CS3 update levels and curves so that they work in 32-bit mode. Please, please, please with sugar on top.

Robert

wrote in message
Robert – the exposure adjustment has nothing to do with converting to 16 bit.

Are you confusing Exposure & Gamma (one of the 4 toning options in the 16/8 bit conversion dialog) with the Exposure adjustment?
ND
Nick_Decker
May 17, 2005
and then adjusting the f/stop

Nope. You should adjust exposure using shutter speed, to avoid differences in depth of field between the various exposures.

I’m looking forward, though, to more in depth discussion/tutorials on this tool. My own experience, thus far, has left me frustrated.
CC
Chris_Cox
May 18, 2005
Again: levels and curves don’t make any sense for floating point data (there is no fixed range). There are lots of things that just don’t make any sense for floating point data (lots of blend modes, adjustments, filters, etc.).

MacBook Pro 16” Mockups 🔥

– in 4 materials (clay versions included)

– 12 scenes

– 48 MacBook Pro 16″ mockups

– 6000 x 4500 px

Related Discussion Topics

Nice and short text about related topics in discussion sections