Blue Bomber
12-19-2010, 09:57 PM
Most people use Photoshop's HDR converter, or programs like Photomatix to make HDRs. I just figured out how to do multiple exposure stacking HDRs, an offshoot of stacking images to reduce noise. This produces a realistic looking image (none of those blown out colors that aren't true HDR).
Let's start with the techniques for reducing noise that got me headed in the right direction. The first way is to take 5 or so shots at the same exposure, open them in PS, then stack (File > Scripts > Load Files Into Stack...), with 100% transparency at the bottom, then 50%, 33%, 25%, 20% etc. That averages out the noise.
The second way is to stack the images, then take all layers and turn them into a Smart Object (select all layers, then right click). Then go to Layer > Smart Objects > Stack Mode > Median. This has the same effect as the first technique. It's quicker to do, but you lose the ability to edit any individual layer (Smart Object turns all layers into one). No big deal if you're just reducing noise.
There's another technique I've found that uses the second noise reduction method, but instead of reducing noise, blends exposures. You can use as little as 2 images (expose for highlights, expose for shadows), too. After you stack the images, but before you turn them into a Smart Object, go to Image > Mode > 16 Bits/Channel. This significantly increases the dynamic range capability of the image. Significantly means beyond the range of the monitor's ability to display it. This is the most important step, as it will allow you much more headroom for adjusting highlights/shadows in the end (similar to a RAW file) .
Next, do the same as the second noise reduction method: select all layers, right click and Convert To Smart Object, then Layer > Smart Objects > Stack Mode. Here's the difference. Instead of Median you'll select Mean, right above it. This will average out the exposures of the layers. Now, you'll have a somewhat flat looking image. You'll want to flatten the Smart Object so you can edit it. Then, you can mess with the exposure of the image however you see fit (Levels, Curves, Exposure, etc.) to get it looking normal. The 16 bit conversion means you can bring up any detail that may still be hidden in highlights or shadow (as long as the detail exists in the original images). Once you've got the image the way you want, change it back to 8 bit (monitor color depth) and do your usual routine of preparing it for the web.
I'll have some examples up soon. :)
Let's start with the techniques for reducing noise that got me headed in the right direction. The first way is to take 5 or so shots at the same exposure, open them in PS, then stack (File > Scripts > Load Files Into Stack...), with 100% transparency at the bottom, then 50%, 33%, 25%, 20% etc. That averages out the noise.
The second way is to stack the images, then take all layers and turn them into a Smart Object (select all layers, then right click). Then go to Layer > Smart Objects > Stack Mode > Median. This has the same effect as the first technique. It's quicker to do, but you lose the ability to edit any individual layer (Smart Object turns all layers into one). No big deal if you're just reducing noise.
There's another technique I've found that uses the second noise reduction method, but instead of reducing noise, blends exposures. You can use as little as 2 images (expose for highlights, expose for shadows), too. After you stack the images, but before you turn them into a Smart Object, go to Image > Mode > 16 Bits/Channel. This significantly increases the dynamic range capability of the image. Significantly means beyond the range of the monitor's ability to display it. This is the most important step, as it will allow you much more headroom for adjusting highlights/shadows in the end (similar to a RAW file) .
Next, do the same as the second noise reduction method: select all layers, right click and Convert To Smart Object, then Layer > Smart Objects > Stack Mode. Here's the difference. Instead of Median you'll select Mean, right above it. This will average out the exposures of the layers. Now, you'll have a somewhat flat looking image. You'll want to flatten the Smart Object so you can edit it. Then, you can mess with the exposure of the image however you see fit (Levels, Curves, Exposure, etc.) to get it looking normal. The 16 bit conversion means you can bring up any detail that may still be hidden in highlights or shadow (as long as the detail exists in the original images). Once you've got the image the way you want, change it back to 8 bit (monitor color depth) and do your usual routine of preparing it for the web.
I'll have some examples up soon. :)