Guys,

Can anyone share how they do your HDR rendering? I tried to do this for some cityscape I have but the images are either blurry and/or doesn't have the "range" i'm aiming for....

Here is basically my workflow:
1. Put camera on tripod
2. Bracket with -/+1stop
3. Use CS2 HDR function

What I would need to know are:
1. How to do this for long-shutter exposures? Any guidelines?
2. What are the critical adjustments that should be set at HDR in CS2
3. What should be the bracketting stops? Sort of rule of thumb to use 0.5EV, 1EV, 1.5EV, etc. based on scene... (I experimented with several -/+EV values, but they seem to render the same result.... Any ideas why?)

I hope the CS2 gurs can demystify this concept for me... :)

Though you can do HDR with 3 images bracketed at -1, 0, and +1 - it would be better to have more images in the range so that there is more data available for the merge. More data would give your more dynamic range.

I normally shoot minimum seven images and maximum of about eleven images bracketed in between -2 and +2. Gives better results than using just three. Downside is that the more images you have in your sequence, the slower it takes for Photoshop to process the entire thing. I usually resize my images to 1024px for test runs. When I find the best settings, only then do I process the full-res images into HDR.

HTH.

Question as well...

Can't you just save multiple TIFF files with different exposure +/- compensations instead of reshooting the same image multiple times? Just wondering...

Like for example a 0EV shot of IMAGE.CRW (or NEF or ORF whatever), then open it in ACR then adjust exposure by +1EV then save as TIFF then do another one at -1EV etc?

Hi Melvin,

First, make sure your scene has enough dynamic range (more than 38 orders of magnitude (log10) to be considered HDRI. More than 13 stops in laymans/photographers term. If your scene doesnt have high dynamic range, no matter how many shots you make, you can't make a low dynamic range (LDR) scene to become HDRI.

1. You need a sturdy tripod with remote release (any slight movement will cause a blur).
2. Bracket +/- 2 stop at least. The more shots the better but the more likely a blur (camera shift) and image (people, grass, etc.) movement occur. So if your cam has autobracketing use it to the max (ideally 5 shots at +/-1 EV). If its windy, a blur in the trees/leaves/grass is inevitable. Avoid scene with moving people.
3. Use RAW and convert to TIFF/JPEG with the exposure, contrast, saturation, brightness set to 0. Set curve to Linear.
4. Merge the converted images to HDR format.
5. Tonemap using Local Adaption or use other software (like Photomatix) or other TMO for tonemapping. (CS2 tonemapping is still very raw. Never had a satisfactory result using it).

Question as well...

Can't you just save multiple TIFF files with different exposure +/- compensations instead of reshooting the same image multiple times? Just wondering...

Like for example a 0EV shot of IMAGE.CRW (or NEF or ORF whatever), then open it in ACR then adjust exposure by +1EV then save as TIFF then do another one at -1EV etc?

David, You can but its not wide enough. You can only extract about 2-3 stops more from RAW. You need more than 2 stops for a real HDRI.

@Nono,

Thanks for the tip! I'll try a 5 or 7 photo sequence. That sounds very logical... :)

How about if the scene keeps on changing within the time you took the 3 or 7 photo sequence (e.g. nitescape with some erratic traffic in the background)? Will this also affect the scene?

Also, what contributes to some blurring in HDR's (considering I have my camera mounted on a very sturdy tripod).

Hi Melvin,

First, make sure your scene has enough dynamic range (more than 38 orders of magnitude (log10) to be considered HDRI. More than 13 stops in laymans/photographers term. If your scene doesnt have high dynamic range, no matter how many shots you make, you can't make a low dynamic range (LDR) scene to become HDRI.

1. You need a sturdy tripod with remote release (any slight movement will cause a blur).
2. Bracket +/- 2 stop at least. The more shots the better but the more likely a blur (camera shift) and image (people, grass, etc.) movement occur. So if your cam has autobracketing use it to the max (ideally 5 shots at +/-1 EV). If its windy, a blur in the trees/leaves/grass is inevitable. Avoid scene with moving people.
3. Use RAW and convert to TIFF/JPEG with the exposure, contrast, saturation, brightness set to 0. Set curve to Linear.
4. Merge the converted images to HDR format.
5. Tonemap using Local Adaption or use other software (like Photomatix) or other TMO for tonemapping. (CS2 tonemapping is still very raw. Never had a satisfactory result using it).

@Mitz,

Thanks a lot!!!

Wow! This is very precise workflow. I'll try to do this. I saw this scene of the whole Metro Manila area from helipad of one of the tallest building here in Makati. I think there is enough range in this. After I perfect this technique, I'll try the 360deg stitching (This is like seeing the whole Metro Manila in one long "panorama"). Cool! I now have a project! :)

I forgot something. To avoid blur, use the "attemp to align image" during HDR merging.

Melvin,
Nice project you're thinking. HDR stitching is still relatively new. Photoshop is not yet capable of stitching HDR. I tried a couple of times using my fave software Hugin. Just imagine stitching multiple 32-bit images and you'll think you need to upgrade your PC hehe.

Mitz: I see, thanks :)

Wanna try one shot Raw for HDR? Visit this site: http://www.nwpphotoforum.com/ubbthreads/information/php/2007_Articles/rjohnston/roman2.php

Wanna try one shot Raw for HDR? Visit this site: http://www.nwpphotoforum.com/ubbthreads/information/php/2007_Articles/rjohnston/roman2.php
for my HDR images, I also use one RAW file but I have a different workflow compared to the site above. Listed below are some of the steps.

1. Using your RAW processing program (I use Rawshooter), create several JPG images in various exposure settings. In my case, I create 5 images whose exposures are -2, -1, 0, +1 and +2. This is similar to the bracketing process indicated in the initial replies but instead of shooting 5 sequential images, you only get one then process the 5 later. This enables me not to shot using a tripod since I am only getting one image. I do not have to worry also if the images are aligned or not since I am only using one image.

2. Once I get my images, I combine them using HDR merge of Photoshop or use Photomatix. I prefer using Photomatix since it is a lot faster and I find its output better than Photoshop.

Here are two sample outputs from this workflow.

http://farm1.static.flickr.com/137/352415516_5ca63bffcc.jpg
I opted for a more surreal tone mapping for this image

http://farm1.static.flickr.com/147/353383251_191c5a113a.jpg
This has a more realistic tone mapping.

i use photomatix pro for HDR convertion instead of photoshop, and use camera RAW instead of multiple exposure from a JPEG file.

i use this technique.. http://backingwinds.blogspot.com/2006/10/how-to-create-professional-hdr-images.html

Due to the limitations inherent in digital cameras, and to a lesser degree in film emulsions, it is not possible to capture the full dynamic range of an image in a single exposure. Even if you capture it in RAW. A RAW file is only 12 bits of linear data, and may therefore still be considered LDR. Until there's a camera that can capture and store/write (32-bit) scene-referred data (HDR, EXR, etc.), you cant have HDR images from a single capture.

One problem with extracting multiple exposure from RAW is the noise in the shadows. You may not notice in web size images or exposure extraction of just +/-2.

Whew! that was a real HDR lesson you got there Mitz!!! thanks...
and in leymans term huh?? Have to gather pieces of my brain though,
everything pumps out after the read (just like what happened in the VR thread) LOL:D

I will try it after i upgraded my reliable PIII, hehehehe!

Whew! that was a real HDR lesson you got there Mitz!!! thanks...
and in leymans term huh?? Have to gather pieces of my brain though,
everything pumps out after the read (just like what happened in the VR thread) LOL:D

I will try it after i upgraded my reliable PIII, hehehehe!


Ricky,

If you want to know the physics and math of HDR acquisition, display, and image based lighting just let me know so i can review my textbooks hehe :D.

Physics and math again??? Oh no!!!
Is it that complicated??? can i bring a calculator?? :D

Physics and math again??? Oh no!!!
Is it that complicated??? can i bring a calculator?? :D

<techie mode on>
I learned physics and math before i learned photography eh. Yes you need a calculator, a scientific one at that.

This is the formula for one of the popular HDR format (Radiance RGBE) (.hdr, .pic).

E=[log2 (max (Rw, Gw, Bw)) +128]

</techie mode off>

Should we not concentrating in taking "good" photographs instead? Hehehe

Thanks for this another HDR workflow. I like the first pic and if I will do an HDR my main objective is to expose everything in the scene.


for my HDR images, I also use one RAW file but I have a different workflow compared to the site above. Listed below are some of the steps.

1. Using your RAW processing program (I use Rawshooter), create several JPG images in various exposure settings. In my case, I create 5 images whose exposures are -2, -1, 0, +1 and +2. This is similar to the bracketing process indicated in the initial replies but instead of shooting 5 sequential images, you only get one then process the 5 later. This enables me not to shot using a tripod since I am only getting one image. I do not have to worry also if the images are aligned or not since I am only using one image.

2. Once I get my images, I combine them using HDR merge of Photoshop or use Photomatix. I prefer using Photomatix since it is a lot faster and I find its output better than Photoshop.

Here are two sample outputs from this workflow.

http://farm1.static.flickr.com/137/352415516_5ca63bffcc.jpg
I opted for a more surreal tone mapping for this image

http://farm1.static.flickr.com/147/353383251_191c5a113a.jpg
This has a more realistic tone mapping.

<techie mode on>
I
E=[log2 (max (Rw, Gw, Bw)) +128]

</techie mode off>

Should we not concentrating in taking "good" photographs instead? Hehehe

@Mitz - Are you serious about this? You actually made use of this... Wow! Ang Lupet (Jologs english translation="WAW! So Cruel!!!"). Hahahaha

@Mitz - Are you serious about this? You actually made use of this... Wow! Ang Lupet (Jologs english translation="WAW! So Cruel!!!"). Hahahaha


Hahaha Melvin.. Yeah im serious. I really did (still doing) a research and study on the subject. Part of my job in computer graphics imaging. But not in photography. You dont need the formula. Just shoot and the let the software do the computations :)

<techie mode on>
I learned physics and math before i learned photography eh. Yes you need a calculator, a scientific one at that.

This is the formula for one of the popular HDR format (Radiance RGBE) (.hdr, .pic).

E=[log2 (max (Rw, Gw, Bw)) +128]
......



im not worthy.. im not worthy.. galing mitz! ur very good at this!