Blending Question

[dwdallam]

I've read Michael's blending essay and several others on PSCS2 HDR. So I thought I'd try something that gave me lots of control and that took about 3 seconds to apply. The result looks very nice, so I know I must be doing something very wrong. In other words, it's too easy and most likely I've giving up much needed information for printing purposes.

1. Open your best exposed image in RAW and adjust the image to that the nicest exposure for the brighter part of the image. Open in PS.
2. Do the same thing and expose for the shadows. Open in PS.
3. Cut and paste the light or dark image into the other.
4. Use the eraser tool at about 20% flow and 100% opacity at the largest brush size you can use for the job and the softest brush setting.
5. Paint over the area you want to to show through to the other image so as to take the best from each image. Repeat strokes until you are satisfied.
5. Have tea.

So this is very easy, you lose no contrast in the image, unlike HDR methods, and there is no masking ect., plus, due to the brush softness and 20% flow, the tonality change is nice and smooth--at least on screen.

So what is going to be the negative with this method? I'm using it only on sunset landscape images.

[feppe]

This is a well-known technique for blending. But I don't see how it can result in a wider dynamic range over processing the same RAW file just once into a 16-bit TIFF file for the following reason:

If you're working with the same RAW source - which is 12-bit from most dSLRs - and process to 16-bit TIFF you don't gain anything by processing the file twice. A 16-bit file can hold all the dynamic range info a 12-bit file has with quite a bit of headroom, so processing twice doesn't introduce any further data. So you should be able to achieve the same results with luminosity masks and curve adjustment layers. Of course, if you're not familiar with those techniques your technique might be easier, but the results are the same.

Someone please correct me if I'm wrong. I must be since there are so many websites which encourage people to process RAW files twice in the matter described by the poster.

[dlashier]

So this is very easy, you lose no contrast in the image, unlike HDR methods, and there is no masking ect., plus, due to the brush softness and 20% flow, the tonality change is nice and smooth--at least on screen.

So what is going to be the negative with this method? I'm using it only on sunset landscape images.
[a href=\"index.php?act=findpost&pid=85773\"][{POST_SNAPBACK}][/a]

The negative is that you can't "unpaint" (unerase), at least easily. You can acccomplish the same result by just applying a layer mask to the top image then spray gun the mask rather than the main image and you have the advantage that you can go both ways (undo by spraying white).

I generally do a combo of both. Starting with a mask created by a blurred version of the light image (or sometimes the dark image) can save you a lot of time, then you just spray to touch up the mask.

Adding a mask is not rocket science, only takes a couple seconds, and gives you the advantage of reversability and the ability to slide in a reconversion without redoing all the work.

- DL

[dlashier]

If you're working with the same RAW source - which is 12-bit from most dSLRs - and process to 16-bit TIFF you don't gain anything by processing the file twice. A 16-bit file can hold all the dynamic range info a 12-bit file has with quite a bit of headroom, so processing twice doesn't introduce any further data. So you should be able to achieve the same results with luminosity masks and curve adjustment layers. Of course, if you're not familiar with those techniques your technique might be easier, but the results are the same.
[a href=\"index.php?act=findpost&pid=85788\"][{POST_SNAPBACK}][/a]

While this is basically true and I sometimes use this method (layering two identical images, each with their own adjustment set and mask), in some cases dual conversions can be easier, primarily because the tone curve work (shadow/highlight rolloff) is already done for you. In other cases (depending on the image) this can actually be a detriment and it's easier to work from the same conversion. In this case I often work with a C1 "linear" image that leaves out the rolloffs and do my own (if desired) in Photoshop. Ultimately, as you point out, the same result can be gotten either way, but sometimes one is easier than the other.

- DL

[feppe]

While this is basically true and I sometimes use this method (layering two identical images, each with their own adjustment set and mask), in some cases dual conversions can be easier, primarily because the tone curve work (shadow/highlight rolloff) is already done for you. In other cases (depending on the image) this can actually be a detriment and it's easier to work from the same conversion. In this case I often work with a C1 "linear" image that leaves out the rolloffs and do my own (if desired) in Photoshop. Ultimately, as you point out, the same result can be gotten either way, but sometimes one is easier than the other.

- DL
[a href=\"index.php?act=findpost&pid=85836\"][{POST_SNAPBACK}][/a]

Thanks for the confirmation that it's pretty much the same - I wasn't sure if I was missing something with the 12/16 bit issue.

I was thinking about the method proposed by the poster and it does, indeed, make sense to just have 2 differently processed RAW files sometimes as it makes things easier.

I usually do blending with images that have extreme highlights and shadows, such as nighttime cityscapes. I work with 2 or more different exposures when I'm blending and slap the darker image on top of the lighter with a black layer mask. Then I just paint in the areas I want to darken and/or use a blurred luminosity mask (which is dead easy to make by dragging the most contrasty RGB channel to selection). So, this method is similar with a RAW file processed twice.

I can see using curves with a 16-bit once-processed file should be easier if the needed changes are global and not as drastic.

[dlashier]

In some cases another reason for going with two raw conversions is that you can WB them differently, for instance in a mixed lighting situation, or to correct blue shadow areas where the sunlit areas are fine.

- DL

[dwdallam]

In some cases another reason for going with two raw conversions is that you can WB them differently, for instance in a mixed lighting situation, or to correct blue shadow areas where the sunlit areas are fine.

- DL
[a href=\"index.php?act=findpost&pid=85868\"][{POST_SNAPBACK}][/a]

Yes, you can do whatever you want in RAW, then open the same image twice, cut and paste into one image.

True enough, masking alloows you to instanly remove anything you have applied, but so too does creating a duplicate layer, and just deleting the one underneath, along with the history brush. I've never really understood masking for 90% of the wrork they say to use them for. About the only reason I can see to use a mask is when your mask (i.e., outline) needs to be perfect. Then you can get down to the nitty gritty by using a mask and painintg very close to teh edge, closer than you could with any selection tool.

[dlashier]

I've never really understood masking for 90% of the wrork they say to use them for. About the only reason I can see to use a mask is when your mask (i.e., outline) needs to be perfect. Then you can get down to the nitty gritty by using a mask and painintg very close to teh edge, closer than you could with any selection tool.
[a href=\"index.php?act=findpost&pid=85912\"][{POST_SNAPBACK}][/a]

Masking is much more powerful and flexible than hard pp'ing, and generally a lot less work to achieve the same effect. Often you can quickly create a suitable mask from the base image, an altered (adjusted) copy of the base image, or from the channel palette. The mask may then be tweaked using any of the normal tools including levels, curves, blur, contrast, hpf, etc., and you can use normal selection tools on the mask to limit these adjustments to particular parts of the mask. The resulting mask can be copied off for use on other layers (including adjustment layers) or image copies or can be converted to/from a selection. If you use layer sets the mask can be applied to the set affecting all members of the set or adjustment layers can be linked to a masked layer and the adjustments will only affect the masked areas. You can toggle the mask on and off to more precisely determine the effects of the mask, etc. etc. etc.

- DL