Ordering Layers

[Kirk Gittings]

I sometimes get into problems with the order of layers. I usually have a mix of adjustment and regular layers including at least: spotting, channel mixer, curves, levels, gradient. What is the logic and correct ordering of these layers and layers in general?

[Daniel_Buck]

I don't think there is any "correct" way to do it, you'll just get different results. At least from what I have seen (been using Photoshop for about 7 years now)

One thing that you must keep in mind, is that photoshop is not floating point compatible (maybe CS3 is. I'm not sure but I doubt it, since it still has different bit depth "modes") So if you push pixels beyond the 255 value you can't bring the highlights back higher up on the stack. So in light of that, setting your white and black point might be best left to the very top of the stack (that's where my final white/black point done) after your curves and other adjustments are done. In general, I usually leave adjustments that treak the highlights till the top of the stack, for just that reason, so I know they don't accidentally clipped somehwere along the line. Programs that are floating point compatable (like Nuke, for instance) it's less critical, because if you push a highlight beyond 255 (or beyond 1, on the 0-1 scale) you can always bring them back unless you do a mathematical clip and force the pixels to clip (not just pushing them higher than 255 or 1, but actually clipping them).

As far as curves and most other adjustment layers (if you have multiple curves) the curves will interact differently with the image depending on how you layer them. (as mid-tones get pushed around, a curve that affects the mid-tones higher up in the stack will not be affecting different pixels) Sometimes it's subtle, sometimes it's dramatic depending on how strong your curves are.

Personally, (for color photographs) I usually do the bulk of the contrast work first, and then work on the colors on top of the contrast layers. That's just my habit.

Hope that helps :-)

[PViapiano]

My personal order from bottom to top is usually Background, Spotting, Sharpening, Levels, Channel Mixer (if needed), Color Correction Curves, Contrast Curves, Hue/Sat, Soft Light Burn/Dodge Layer.

Works for me...and I've tried re-ordering to see different effects, but this is the one I've stayed with.

[Kirk Gittings]

Thanks, I should have said this is a b&w workflow that I am talking about.

[PViapiano]

Kirk...same thing. I don't include the layers I don't need...

[JW Dewdney]

I just do it the easy way and flatten after every single step...!

[jetcode]

I just do it the easy way and flatten after every single step...!

ouch! so much for balancing the effects of each layer in the processing stack!

[jetcode]

Kirk...same thing. I don't include the layers I don't need...

Curious if you archive the PSD (layered stack) version of each image you process.

[Henry Ambrose]

I accomplish what I want with a layer or layers then flatten as soon as possible. I never use layers for spotting - spotting belongs on the background. If its a typical sorta soft flatbed scan I also do a gentle global sharpen on the background. Things that are for sure a part of basic correction back to what I imagine is "on the film" or "basic reality" belong on the background or flattened to the background after being made. Then I save the file.

I really like having a compact file. Even on a powerful computer there's a big speed and working difference in a single layer 200MB file and the same image with 5-7 layers and over 1GB size. Also a clarity of purpose shows up when there are few layers. There is no question about what am I really doing and how does it effect the 4th layer down the stack.

And just because we can have layers doesn't mean we have to or need to every time. Kinda like camera movements - lots of time shooting straight on is the best solution. That the camera is tied in knots does not mean the picture will be better. Same for layers.

[bob carnie]

Kirk
I do not have an answer, but I am testing various issues with capture to Black and White on our Lambda and would be willing to share my results but this will take the winter to complete them all.
One test was to do a stacking test just as you are questioning, another to compare the various ways of dodging and burning. using different tools with CS3.
Resetting aim points for 21 step grey scale tests is also a very important area for me which we are playing around with.
I would be interested to follow this thread as well as hear your viewpoint after other folks from this forum jump in.
One thing I would suggest in any testing is to use the highest gloss material that you can use. I have found that using the Harmon Fibre Base on our Lambda, artifacts are more clearly evident than when using a rag paper on our Epsons.
Also pushing the test size up to 40 inch test strips are very helpful.
So a quick test IMO would be for you to take any image stack the layers the way you normally would , flatten and name normal. go back to the psd file change the order save and rename, and when you have finished all the different changes/renamed them and saved , print test strips at high magnification . In my shop we leave sharpening for the lambda and do not do it in PS but I am not sure of your work flow, but I would consider this aspect.
You should either see absolutely no difference or you see some difference to make your decision. I do not know what you will see but I think its a good test to try on an existing image you already have finished and like.

[Daniel_Buck]

I'll usually save the file with all the layers intact. Usually that's 5 layers (in this order): image, base Curves, "burn" curves, "dodge" curves (both with layer masks so I can edit them later) and then my levels on top. For a B&W image, it's rare for me to have more than those 4 additional layers on top of the original image (if the image is a B&W scan). If I need to do some simple clean up some dust hits, I'll do it to the original layer. If there are some scratches or large dust hits in areas that have detail and require more work, I'll leave a duplicate the layer underneath so that I have my original layer under there just in case I need it.

I've evolved (as has photoshop!) so that everything I do (aside from dust hits) can be un-done, or tweaked if I need to. I like this method, because often times I'll come back to the image the next morning, and find myself feeling different about how I've set the contrast. (especially if I had been working on the image the night before after a long day at work!)

Drive space is cheap, and I don't shoot alot (only a few shots on a weekend, and I usually take one or two of them for a high res scan, unless I end up with more great shots)

[Kirk Gittings]

Resetting aim points for 21 step grey scale tests is also a very important area for me which we are playing around with.

What does this mean Bob?

FWIW

For b&w I archive a approx 300MB RGB file with all curves it runs about 1GB, which I find a manageable size to work on. In addition to the above I have a sharpening layer. I use ImagePrint which automatically flattens the image and decently downreses the image. That way I only archive the original scan and one large working file.

[bob carnie]

Kirk
On the lambda , before any work can be done each material is calibrated to a 21step grey scale that is based inside the software driving the machine and then the rgb curves are balanced in to the papers specs.
The harmon paper I am referring to is real black and white silver gelatin paper*gallerie grade 4 with an extended red sensitivity.This is not an inkjet paper. Since this paper is basically an panchromatic emulsion , we can run it in RGB and process the paper as normal with the safelights out. After processing it is just the same characteristics as gallery 4 and tones just the same.
Every manufacturer supplys aim point for Dmin and Dmax and from those point the red green blue curves are plotted in 19steps between the Dmin and Dmax.
Each material has different aim points for upper and lower scales as well the rgb aims in both are never the same.
Fuji FCA and Kodak Endura have similar *slightly different aim points* for Dmin and Dmax.
Metal Paper has a much higher Dmin aim point and the fibre paper has very low Dmin aim points.
Most all capture devices and photographers set up their files to print on inkjet or RA4 output where the dmin is at a much higher point than the paper I am using. So some devices and PS workers are sending files that are meant to have the highlights print on those respective dmin aim points. but for my paper the whites are much brighter and there are problems with certain capture devices*phase* but not with DSLR or our Scans.
We are finding that by raising the Dmin aim point for the paper we are able to match a better interpertation of the clients work.
When I first started this process we were using Agfa Classic in the Lambda and the Dmin aim points were higher and we had no problems with phase files, now with harmons paper we are being forced to change these points to place the highlights properly.
Sounds confusing, it is and that is what I mean about playing with the 21 step wedge.
Its somewhat like placing a full bodied negative that prints lovely on a grade 1 paper onto a grade 4 emulsion, highlights blow out and all kinds of problems.
We think that phase captures at a much higher dynamic range and those workers are placing the highlights in regions that for Fuji FCA RA4 or inkjet there is no problem but when the put the same file on my paper the dmin aim values are much lower and the image is causing artifacts.
It is extremely difficult to tell clients to adjust their processing of the captured files to a lower contrast there fore we are trying to solve the problem by raising our Dmin.
This only has shown its ugly head with phase capture and not with DSlr or scans that we do. go figure.
We only keep the Original File, Working PSD and Lambda PPM. Once a year clients are called and all files are returned and if not wanted we dump. The lambda has a wonderful sharpening feature that we apply at the very end and is not part of the PS workflow unless local areas need sharpening.
Beyond this explanation , I would defer to Durst technicians or Harmon technicians regarding the problem that I am having, since there are currently only 5 labs that I am aware of in the world doing this process, its kind of hard to problem solve ,as we do not talk to each other.
best regards
Bob


What does this mean Bob?

FWIW

For b&w I archive a approx 300MB RGB file with all curves it runs about 1GB, which I find a manageable size to work on. In addition to the above I have a sharpening layer. I use ImagePrint which automatically flattens the image and decently downreses the image. That way I only archive the original scan and one large working file.

[Jack Flesher]

Bottom to top, usually: Base image, clean/spot, detail-sharpen, local contrast, White-Point adjustment, global contrast (curves), color tweak (hue/sat, curves or both), dodge/burn, output tone curve. One of the things you can do to simplfy the stack when you feel a set of adjustments should stay together as a logical group, is to group them into one "smart" object. That set will remain intact and act as a single layer as you manipulate the others, keeping the additional layer "balancing" steps to a minimum.

Cheers,