|
kriso9tails
Mar 6, 2008, 2:03 AM
Post #1 of 4
(1401 views)
Shortcut
Registered: Jul 1, 2001
Posts: 7772
|
Got any? Post 'em up. edit: it took me five attempts to type 'post'.
(This post was edited by kriso9tails on Mar 8, 2008, 9:23 PM)
|
|
|
|
|
kriso9tails
Mar 6, 2008, 2:04 AM
Post #2 of 4
(1396 views)
Shortcut
Registered: Jul 1, 2001
Posts: 7772
|
First off... excuses: i) This is the only non-work shot on my computer and it's just an incomplete rough concept draft... don't make fun of my sloppy/ lazy cut and pasting (around the hair, and most of a draw and part of a sling are just cut off) ii) I'm doing this 'cause I was working, but have now decided to just run down the clock at the end of a slow work day: if the screenshots are an odd size for you, I really don't care and despite doing this... I'm too lazy to make it better. Now: Alternate method of contrast adjustment through overlaying black There are a lot of ways to vary the same basic idea, but this concept is a useful way to raise contrast by overlaying black. Why do this instead of levels or curves? It's easier to control and it will be less harmful (or even harmless) to your histogram. On the image below, the climber appears too dark, but in actuality there's a lot of detail that's there, just not really coming through. Unfortunately, it's taken from a jpeg where the file integrity just isn't that great. If I try to use curves or levels I'm going to get unacceptable amount of combing, which I just can't afford here, so instead I'm just going to bring up all of the values and layer the blacks back in. First, I'm going to lighten the image with a shadow/ highlight adjustment (Image --> Adjustments --> Shadow/ Highlight). Shadow/ Highlight isn't always the best solution, but it can be pretty damn useful (I just wish Adobe would make it an adjustment layer option). In this case I'm only going to adjust the 'Amount' and 'Tonal Width' sliders until I can clearly see all the details I want without blowing out any of the highlights (that weren't blown out already). Actually, I'm just going to leave it at the default values (50 and 50) 'cause once more... I'm lazy. Alright, next I'm going to head into the channels tab. It's a little unclear in the screenshot, but what I'm doing is duplicating the red channel, or rather, making an alpha channel from it. Why red? The red channel best defines contrast, green - midtones, blue - uh... noise. Now, to get the dark values that I need I'm going to make a curves adjustment to my new alpha channel. For the most part I'm just going to drag the bottom left anchor point on the density 'curve' over to the right until I have as much black as i want without killing any of the detail I want. I don't really want to affect the upper half of my curve too much. I was going to say "Now I'm just going to copy the alpha channel into a new layer" but for those that don't know how to do that, the next three screenshots deal with that. First, go back to the layers tab and create a new top layer. Fill this layer completely with black. Second, go back to the channels tab and select your alpha channel (command click the channel in the palette [cntrl click for un-macs]). Third and last, go back to the layers tab and on your new black layer, fill the selection with white. Almost done. On the new layer that I pasted the alpha channel into I'm going to go to the blend mode drop down menu and select 'Soft Light' and voila! (If you're not jumping up and down shouting uncontrollably at the top of your lungs at this point, I feel as if you're not as excited about this as you should be). If you get to the end and find that the effect is too strong then just reduce the layer opacity. If there's parts of the image that you don't want it applied to then simply mask them (or you could erase those parts from the top layer, but erasing is so... permanent). Yeah, so that's that and hey... I can even book two minutes of overtime now.
(This post was edited by kriso9tails on Mar 7, 2008, 9:02 PM)
|
|
|
|
|
kriso9tails
Mar 6, 2008, 7:20 AM
Post #3 of 4
(1353 views)
Shortcut
Registered: Jul 1, 2001
Posts: 7772
|
Old 'trick' and much simpler: Optimized Image Enlargement (without genuine fractals). If you want to size a photographic image up (or any raster image for that matter), Photoshop is going to have to resample pixels, or to put it ineloquently (and maybe a little inaccurately), make a mathematical guess what pixels to add based on the surrounding pixels. The bigger your enlargement is the more pixels Photoshop has to add thus the more it has to guess. The solution is simple then; make multiple, smaller enlargements in succession. In the Image Size dialog box, change the resampling method to bicubic smoother. Next, change the measurement units for width and height from inches to percent and enlarge by 110%. Just keep repeating this until your image reaches the size you want. Obviously this can't add detail or clarity where there was none to begin with, but it will prevent your image from losing any through bad resampling.
(This post was edited by kriso9tails on Mar 10, 2008, 8:32 PM)
|
|
|
|
|
fiend
Mar 10, 2008, 5:43 PM
Post #4 of 4
(1271 views)
Shortcut
Registered: May 25, 2001
Posts: 3669
|
Thank you for sharing, kind sir. I found your delivery concise and relevant, while still remaining detailed enough to help me through the process. I particularly enjoyed the use of real-world examples and screenshots to illustrate your points. I just have one question... what if the monkey doesn't feel comfortable working the controls on the blender? Seriously, I have nothing to share but would like to see more.
(This post was edited by fiend on Mar 10, 2008, 6:43 PM)
|
|
|
|
|
|