Touching Up A 3d Image In Post
No matter how much I may work a 3d piece to death, it's never finished until I spend an hour on it in Post. "Fixing in Post" is just a fancy way of saying you don't have to skills/time/patience/means to deal with an element during the piece's creation, and instead you address it post-process in a more hands-on program, like Photoshop.
It used to be that most 3d stills you'd see that contained anything pretty looking had been doctored in post, but nowadays more and more "post-process processes" are becoming integrated with the 3d Package. "Doctoring a still in Photoshop" is becoming less and less necessary. But in case your 3d Package doesn't include some of these processes, here's a tutorial on how to achieve some basic Post effects in Photoshop.
This Tutorial covers: Backgrounds(Alpha Channels), Glows, Depth-of-Field and Blurs, Noise and Color Adjustment. Along the way we'll explore Channels, Selection Tools, Masking, Blending Modes, Quick Masking, Adjustment Layers, and more.
NOTE/DISCLAIMER: This tutorial is intended for people familiar but novices with Photoshop, although it contains enough tricks and tips to make just about anyone feel like they've learned something. The example piece I'm using is going to jam in every trick to the fullest, so remember this as the 'Dante's Peak' of examples (If it can be added, it's going to be). I can't stress enough how most images don't need ALL of the following edits to their full effect and you should use your own discretion. Also, please note that there are a thousand ways to do anything in Photoshop, and my way may or may not necessarily be the best way. Learn to develop your own technique.
Here's the piece we're starting with, you can grab a .TGA of it here:
Here's what we'll end up with, Photoshopped like crazy:
Before doing the tutorial
Render out your 3d piece to a bitmap format, preferably as a targa file (.tga). The Targa file format is high quality and usually incorporates an alpha channel (see below). If you don't have the option of saving an alpha channel, then click here for steps on making your own.
Alpha Channels are basically saved selections. When you save out an alpha channel from a 3d program, a small black and white image is created that can be used to separate your object from the negative (surrounding) space. Usually the alpha channel is found as "alpha" under the Channels Palette in Photoshop, but I've seen a few times where a 3d application will save the "Alpha Channel" as it's own black and white bitmap file.
Now there are a lot of different things you can do to an image in Post. You can either click on one of the following steps to skip to it, or keep clicking next to follow this tutorial in order.
Adding The Background
To start off, open up your 3D image in Photoshop. If you want to follow along using this exact file, then you can get it here. It's a good habit to right away duplicate your base layer, so you're always working on top of a copy of the original (left click on layer, choose duplicate layer). Then, lock and hide your base layer, just to be safe.
Right away the solid-light-blue-negative-space attracts the eye, so that's got to be replaced by something a little more divided. If you're ultracool then you'll have created the background in 3d and can skip this step, but if your not ultracool then you can do things this way.
Choosing an background image:
The background image is only meant to take up negative space, meaning it should be something that does not attract attention and the more it blends in, the better. For this reason, a blurrier diffused image will be used.
I'm going to choose this picture by 3dTotal's WebGuru Tom for my background, simply because it has elements that work for my piece and Tom will probably get a kick outa this. Now, as it is, this image won't work so well as a background...
But after heavy doctoring it'll be perfect! By cropping, rotating the image, blurring, and adjusting the levels I was able to create something that sufficiently broke up the background area for me. (I planned ahead a little with this, so trust me, it should look OK in the end). You can grab this image to work with here.
Creating a background will be a little different in every scenario. You can use just about anything to create something that blend in and no one will notice. In fact, the less it's noticed, the better.
Once you've got your background ready, then copy and paste it into a new layer on your 3d piece. Now open your Channels Palette and hold the Control button while left clicking on the "Alpha" or "Alpha 1" Channel.
This loads the channel as a selection. (Ctrl+click works for layers and masks as well, by the way).
You may need to invert your selection, depending on whether or not your alpha channel had the background as black or white. Ctrl+click will select whatever is white in the channel, which you can see (in the tiny thumbnail in the picture just above) is the foreground, not the background.
If it's necessary to select the invert, simple hold Control and Shift while you tap the "I" key. Regardless, you want a selection of what your background will be.
Now, click back on the Layers Palette and click on your layer with the "background" (be careful not to mix this up with the "Background" layer, which is Photoshop's default name for the lowest layer in your stack of layers) By clicking the little mask layer icon, you'll automatically create a mask over the layer in the shape of your alpha channel.
This is what you'd end up with.
It doesn't look like much, but it's a start. Now's a good time to invest in some good habits. Rename your layers so you won't get confused later on. I renamed my base layer as "original", my base copy as "work copy" and my masked 'background' image as "negative space". You may need to unlock the bottom layer to change it's name.
It's also a good idea to unlink the mask from the layer. Do this by clicking the little link/unlink symbol. After you unlink the mask, click on the little layer thumbnail to manipulate just the layer. Click on the mask thumbnail to manipulate just the mask. You would link the mask and layer when you want to manipulate them together.
The final good habit to get into is to save. A lot. Save this file as a .psd to preserve all the channels and layers.
Now that the background is in place, we need to adjust it so that it looks like it fits better. The original negative space was a light blue that I had selected to match the image, so by lowering the layer opacity I can let a little bit of that blue shine through.
By changing the blending mode of the layer to overlay, I was able to increase the negative space's illuminance.
I brightened the opacity again too, finally settling on 85%. This more luminous feel will become helpful later on when we begin to work with the backlighting glow.
For now the background is about done, and we're ready to move on to the next step, Glows and Highlights.
Glows and Highlights
Glows and highlights can really make your image stand out. But they're very easy to overdue, and in fact, this example probably has too many, but that's OK.
Here's where we left off, with the background in place:
The image is naturally backlit, so simulating the light coming from the background will be the main task here. To start this, we need some selections.
Open your Channels Palette and Ctrl+leftclick on the Alpha Channel to make it a selection. Next go to your Layers Palette and create a new layer. With the new layer selected, click edit>fill and choose white, opacity 100%. We didn't NEED to fill this with white, but it helps give an idea for where this is heading.
Without deselecting, choose select>feather (shortcut ctrl+alt+D). Use a setting of 15 pixels and hit OK. Feathering the selection partially selects pixels around the start selection, and 15 pixels is just a range I think looks good in this particular case. Go up to edit>fill and choose white and 100%. Do that twice, so you end up with something like this:
Sure it looks glorious, but it hurts the eyes to look at too much, and it hides the pretty background stuff we made.
So first things first, we need to lower the layer opacity a little. I used 72%, which isn't much, but we're going to take the backlight down even further.
Deselect everything and click on your Layers Palette. Add a mask to the layer by on the glowy layer and clicking the "Add Mask" button at the bottom of the palette.
The mask should immediately be created and selected, so go right away to edit>fill and this time choose 100% Black. This hides the Glowly Layer we created and allows us to reveal it bit by bit.
Still working with the mask, change your foreground brush color to white, and select the airbrush. Using a fairly large brush
begin to "unmask" parts of the layer where the glow would most likely appear. I used a brush opacity setting of 40%, but that's personal preference. Here's an example of what I had after a minute or so.
It's a good idea to click once and evaluate, then click again and evaluate, instead of just clicking and dragging. This will allow you to decide whether or not the glow really works where you let it shine through.
This image shows my individual clicks with the airbrush tool in red, the bigger the area of red, the more times I clicks and thus, the more glow I let through.
Continue letting more and more glow through, while working around the elements in the background. For the windows, I only let the glow come through the bottom parts, since the light was coming from the top of the picture and wouldn't be able to creep around the top inside the of the window as well. Here's what I finished with
The brightest area is definitely in the at the center towards the top and the far right window. I did this on purpose to balance out the next glow we're going to work on.
It's a good idea to rename your glowy layer to "glowly backlight area" or something more creative. Save too.
You can't have a glow-in-the-dark lego skull without having it glow, but this is going to be a little trickier than the backlighting. Why? Well, two reasons:
1: We don't already have a selection of the skull
2: It's a much smaller area and has important details (the face). Those details need to stay pronounced while the glow needs to mimic what real glow-in-the-dark objects look like.
Problem 1 is not difficult to take care of. And while the magnetic lasso tool would work really well at first glance (due to the high contrast of the area), it's not the best tool to use.
If you look at real glow-in-the-dark objects they cast a little illumination, but most of the glow is within the object itself, like an inner glow. If we use the magnetic lasso we would then have to shrink and then feather the selection to prepare for this inner glow, and then we would have to repeat this for a smaller illuminating outerglow. It's easier and quicker to just use something called Quickmask.
Quickmask mode lets you literally "paint a selection". You can use brushes, gradients, filters, even adjustment effects to create whatever selection you need. But it is a little tricky to get the hang of at first.
To shift into Quickmask mode, click the little right button just below your "foreground/background color box things". While in this mode, your "foreground/background color box things" will be grayscale, similar to how it looks while working on a layermask. But instead of masking or revealing a layer, if you paint in quickmask it'll highlight the area on your picture, defaultly with a bright red.
At anytime you can shift back into regular editing mode by clicking the little button on the left, just next to the quickmask button. This turns anything you "painted" into a marquee.*
*Even though you won't see it when you switch back to regular mode, Quickmask will create selections containing partially
selected pixels. Like any selection, Photoshop will only display a rough estimate of pixels selected (by showing those that are "mostly" selected).
Keep in mind, with Quickmask, the more red, or highlighted an object is, the more it will be selected. The less red it is, the less it will be selected.
Using Quickmask and a fairly large airbrush, change your opacity to something really light, like 10%, and try and paint something like the below:
See how there's only a tint of red, that means there will only be a little bit of a selection along the outside, barely enough to be noticeable. That's what we want. Switch to a smaller brush size and up the opacity to something like 80% and paint along the inside of the skull, clicking much more this time. Try and get something similar to the image below:
That deep red tint means that those pixels will be completely selected when we switch back to regular mode. There's one problem, the edge of that pole that overlaps the skull shouldn't be red at all, since it's way in front and wouldn't pick up any of the glow. Use the Polygonal Lasso to select the red area of the pole.
Simply hit delete to get rid of any red overlapping the pole.
Now it sure would be a shame click somewhere and lose our pretty selection after all that. Click on Select>Save Selection to make sure you'll always have it.
If you look close you'll see a field marked "channel". That's right, it saves the selection as a channel and you can open your channels palette and see it right there.
Anyway, saving selections is just another good habit to get into. Let's move on to creating the glow already. Create a new layer and rename it "skullglow". Next we need a color to base the glow around. Start with a color sample (using the eyedropper) from the skull and then use the Color Palette to adjust the color until it resembles that glow-in-the-dark-green-glow-color.
Here's what I settled on:
Now if you lost the skull selection then switch over to your Channels Palette and Ctrl+click the "skullselection" channel to restore it, then click on the skullglow layer and select edit>fill. This time choose foreground, 100%.
The glow itself looks pretty good, but what about the face.
That's actually really easy to fix, just change the layer blending mode to "overlay"
It doesn't look that bad actually, and the image is just about ready for the next step, which is Depth-of-Field and Blurs. Here's a final image of where we are so far.
Depth-of-Field and Blurring
Depth-of-field, in traditional photography terms, is the range of the focus on a lens. And, technically, in every photo it's always physically there, but could either be a narrow or a wide depth-of-field. A close-up of a flower usually has a narrow depth-of-field. A huge landscape shot from far away usually has a wide depth-of-field.
In the 3d world, the term is used to really only to describe the close-up effect. While you should always make some effort to blur out less-important parts a little, using depth-of-field effects can really make your main focus obvious.
Nowadays most 3d Packages come equipped with a depth-of-field filter or effect already, but here's one easy way to fake it.
Here's where we left off:
When thinking about depth-of-field you need to think about focus, this is the target area that I want to remain focused, everything in red needs to be blurred at least a little, if not drastically.
To start off the blurring we need to duplicate our work copy layer (name the duplicate "blur copy") and then turn off every other layer except the "blur copy" layer and the "work copy" layers.
Select the "blur copy" layer and use Filter>Gaussian Blur on it. I used a radius of about 6.7.Yes, this Gaussian Blur does blur everything. But we're going to use masks to minimize the effect in our focus areas.
After clicking OK then apply a mask to the "blur copy" layer.
By painting black on the mask, we can hide the effects of the blur and allow the crisp "work copy" layer to show through. This really is a matter of preference but its important to keep in mind the distance areas are from your focal point. I started by using the airbrush at 40% to reveal my focus area.
At this point you need to visualize your 3d scene.Things that are either really in front or really behind the focal point will become blurred out, here's how I visualized the image.
After I finished painting the mask, here's how it looked, it's OK if yours doesn't look exactly the same, just remember, when working on the mask in this case, what is white will be completely focused (from the layer underneath) and what's black will be completely blurred. Anything grayish will be a mix somewhere in-between.
Use the bracket keys "[" and "]" to change brush sizes. Use Shift+brackets to change between harder and softer brushes. Use the numpad (0-9) to change the brush opacity.
You can see that I didn't follow my "visualization" exactly, but it's pretty close. I focused or blurred additionally for design purposes too.
Here's what the mask would look like in the actual image. There are plenty of other ways to blur parts of your image, but this way maintains a focused version underneath it all too.
After unhiding all the other layers, we can see our picture is starting to get that "Photoshop't to death" feel. That's OK, this is the "Dante's Peak" of examples: something we're using to shove as many different possibilities as we can imagine into it at once. We're now ready to move on to the next step, Noise!
One thing that always seems to throw off the suspension of disbelief in a 3d piece is it's prestineness. By adding noise, scratches and other real-life elements into the picture we can subtly enhance it. It looks dern cool too.
Here's where we are so far:
Adding noise is pretty much a guessing game.
You could just add a noise filter on a separate layer and lower that layer's opacity, but the result would be rather bland overall, and there are simple ways to vary the noise effects.
You could also use any number of grunge maps, noise maps or other templates (there are some nice ones on 3dTotal's texture CD!), but what if you don't have any available to you?
This tutorial will go through a simple way of creating a varied noise effect just by using Photoshop's native filters.
Start by creating a new layer and naming it "noise". Hit the "D" key to default your colors. Now apply a Filter>Render>Clouds filter to the new layer. Right away apply a difference clouds filter (Filters>Render>Difference Clouds) to the layer and keep applying it until you end up with something you like. Clouds and Difference Clouds create a random tileable pattern everytime, so keep trying it until you end up with a result that's high in contrast and has several obvious "veins", like below.
Now this part will be different for everyone, but open Image>Adjust>Levels. Drag the middle (gray) slider a bit to the left, making the veins a lot more pronounced.
Now you can apply the noise filter, I used the settings below:
Amount: 21.33, Uniform, Monochromatic
Now, here comes the trick. Lighten the layer opacity to about 10-15% and change the blending mode to screen You should end up with something like this:
You can sort of see the veins, but they don't detract from the image. You can further enhance or diminish the effect to taste by using image>adjust>brightness contrast on the layer. Here's how mine ended up. It looks a little too patchy, but that will be fixed up later.
Next: Color Adjustment and Final Touches!!
Color Adjustment and Final Touches
An over-the-top 3d retouching example wouldn't be complete without some last minute color adjustment. This process is really easy and can be used for just about anything, not just retouching 3d pieces in post. Here's the image where we last left off:
Ever since Photoshop 6, Adobe has included these neat little things called "adjustment layers". Basically it's any of the "image>adjust>" effects that can be used as layers themselves, effecting anything below them and maintaining editbility. That's what we're going to use.*
*If you don't have Photoshop 6 or above, you can do this, but you need to save a flattened version of your file and use "Image>Adjust>" on that. Otherwise you won't affect all of the layers collectively.
Click on the small half circle icon on the bottom of your Layers Pallette. This opens a menu of various adjustment effects. Choose "Color Balance"
I wanted the image to have an overall more "blueish" tint. To achieve this I used the following settings:
This resulted in a much more "cold" and brooding feel, which is exactly what I wanted. Using subtle tint of color in your work can affect the overall feel. Follow the old color theory idioms: Reds are warm, energetic, and pop out. Blues are cold, slow and recede.
Other touch ups:
After all is said and done, I played around with the noise layer a little bit and I also applied adjustment layers to touch up contrast and saturation a little. Keep in mind working on a piece in Post is half technique and half personal preference.
Here's the finished image.
versus the original:
And here's a view of the final Layers and Channels Pallettes:
There are other effects we can apply to the picture, but for the most part it's about done. It may look a little too glitzy with all the different edits to it applied at once, but no one says you have to use all of them. Pick and choose what you think looks best in your image and if you can, try and use the 3d Package itself to calculate effects (especially backlights and depth-of-field effects). Don't feel the need to go as overboard as this example, being subtle can go a long way. Look at the Gallery and see if you can spot any possible post effects; on the best images you won't be able to tell if the effect was rendered or created in post.
I hope you found this tutorial to be useful/informative.