home about gallery publications education links contact

..::photoshop CS tips for 3D::..

Not just limited to 2D illustration, Photoshop can seriously aid the 3D artist in his or her day to day routine. With the new panoramic stitching tool, the 3D artist can save time by letting Photoshop CS do the majority of the back-breaking work, just leaving a little clean-up. The new non-destructive pixel aspect viewer tool helps the artist view the image how it is supposed to be seen on the target screen, saving time by not having to pipe the images / animation to the screen for testing and advanced colour correction helps tweak the CG elements to match the background plate, increasing productivity which will please not only the artist, but the client as well! Here we run down some of the more common tools and processes that I think the 3D artist will benefit from the most…

Ageing the Image
The ability to load in Filmstrips (which can be outputted from Premiere (for example) if your 3D package does not support it) is a feature which is often underused. One great example is the ability to add scratches to the “film” by creating marks right down the strip to give the impression of aged footage. In addition to this, try dropping the levels of one channel in a vertical slice of the filmstrip to suggest weathering or even use a grunge brush to suggest damaged film.
Degrading the Image
Unless you want your renders completely crisp, to remove that CG feel try degrading the render a little. This can normally be done in post in your 3D application, but Photoshop handles it a lot nicer. Double the size of the image, add a little Gaussian Blur to the render… 0.5 to 1 should suffice (depending on the size of the image) and add a slight amount of Gaussian Noise to break up any steep contrast and bring the size back down to its original setting.
Fixing Irregular Panoramic Exposures
If you have taken a panorama of, say, a cloud formation, and you have forgotten to lock the exposure while taking the image, all is not lost. By using the new Match Colour feature or by tweaking the levels and/or brightness and contrast you can match up the exposure from one image onto the next. For this to work effectively, it is best to work in layers in one large canvas and not on individual images so you can match the shading in any overlapping areas easily.
Stitching Panorama Images Together
The resulting images for the panorama can then be either manually placed and distorted to fit, or created using Photoshop CS’s new Photomerge feature which will overlap and stitch the images together. However, this feature is not completely perfect and some fading between images may be apparent. Therefore you may need to manually cut and paste (or use the Snapshot tool) some of the source images into the new (flattened) image to remove any fading.
Tiling a Panoramic Image
With our panorama completed, we need to get the image to tile horizontally to remove the seam when both sides meet up. To do this, use the Offset filter set to half the image size and 0 Vertical offset so that the edges of the image are centred. Next, paint out (and colour correct if necessary) the seam by using a combination of the Snapshot tool and Healing brush. Offset the image back using a negative amount if desired.
Enlarge Screenshot
Removing a Cylindrical Map Pinch
As this panorama is (more than likely) going to be applied to a hemispherical object which encompasses the entire scene, the panorama will be applied using cylindrical mapping which will cause a pinch at the top of the hemisphere. To remove this, apply a Polar Co-Ordinates filter set to Rectangular to Polar and paint out the pinch in the middle of the resulting image. Finally, apply a Polar Co-Ordinates filter set to Polar to Rectangular to put the image back and apply the resulting panorama to the hemisphere.
Grain Matching
Currently, Photoshop’s grain management is limited to a few simple settings such as the amount and intensity. Chances are that the grain on the backplate you are trying to match is of a different size and may have more red, blue or green. Therefore add a new layer (masked using the opacity of the CG layer) and flood fill it a mid-grey. Change the layer blending type to Overlay or Soft Light (some slight tinting will occur which should be rectified) and apply the Film Grain (or Noise) filter at the right intensity. Now you can adjust the levels for the noise to make it more red, blue or green and also resize it.
Photoshop as a Compositor
If you don’t have access to any additional compositing software such as Adobe AfterFX or Discreet Combustion, Photoshop can be used as a “poor man’s” compositor. The entire animation sequence can be composited by recording an action, finishing off with a flattened image. Perform a test run on a different frame to ensure everything runs smoothly; if not, insert a stop or two in the Action sequence so you can re-record the offending item. Finally, batch process the entire sequence, not forgetting to output to a new folder as not to overwrite the originals!
Enlarge Screenshot Colour Correction
In additional to compositing your animation onto backplates and adding foreground plates, Photoshop can be used to colour correct an entire sequence by simple colour replacement. By using the Replace Colour tool, and entire sequence’s colours can be amended should the client change their mind on the colour scheme. You can also adjust the entire range, adding additional contrast and slight desaturation to match a background layer, record it in an action and perform it on the whole sequence.
  Rendering to a Specific Print Size
Even though a lot of 3D applications will render out to 300 dpi, getting the final pixel size for a specific print canvas is difficult to determine as there is (often) no physical canvas size (in mm for example). Therefore create a blank canvas in Photoshop at 300 dpi to the correct dimensions (including bleed if necessary) and note down the width and height pixel settings to input into your 3d program. Even if you have to render out at 72 dpi you can still drop the image directly into the 300 dpi canvas without any tweaking by simple copying and pasting straight in.
  Pixel Aspect Ratios
Photoshop CS now includes the ability to display different pixel aspects which is ideal for the 3D artist who renders for production. Although this is simply a temporary viewing tool (it is not saved with the image), it does give a good interpretation of what the final image will look like on-screen. This is ideal when using Photoshop as a compositor as we can constantly check the final image output instead of having to second-guess the result.
Enlarge Screenshot Toon and Edge effects
If your 3D application does not have a toon shader, we can create flat toon-shaded images by amending the image in Photoshop. Render out a standard shaded image from your 3D application and apply a Poster Edges filter. This works to a certain degree, but if you need finer detailed lines, render out another pass with alternating colours for each object and set them to 100% self-illumination. Drop this render into our composite, set its layer blending type to Multiply and apply a Find Edges filter to “draw out” the lines (you will need to desaturate the result).
  32bit Targa Problems
Some 3D applications allow you to render out 32 bit TGA’s (images which contain an alpha channel) which makes for easy compositing. Photoshop 7 will took this 32 bit image and make the transparent alpha parts of the image completely transparent and ditched the alpha channel so if you have rendered onto a backplate (which is not included in the alpha channel) this will be lost. Thankfully, this has been rectified in Photoshop CS so we can now render out 32 bit TGA files instead of having to use a different format or rendering off two images (a standard render and an 8 bit Alpha image)
  Cleaning up Baked Textures
If you have baked out textures for real-time 3D (or simply to save render time), you will need to clean up some of the resulting textures due to some artefacts being present. Finer detail of original (now baked) textures may be occluded so rebuild these in Photoshop by working on a larger version of the baked texture and resizing it down, or bake out a larger version, amend as necessary and resize the tweaked textures to bring them down to a manageable size for the game engine to handle.
  Texture Degradation
No surface is perfect; depending on its location and usage, surfaces become grimy and damaged, especially painted metal. To create a map to reflect this, set your brush’s width setting to pressure (if you are using a Wacom or equivalent tablet) and add scratches. Use a large patterned brush to add large chips in paintwork. Export the resulting painted black and white mask as a bitmap so it can be used as a bump map in your 3D program and also to control reflections and specularity (or to simply mix two different materials together – one with our original (slightly dirtied) surface, and the other with the bare worn-down metal.
  Removing Colour Tints
When using atmospheric effect such as fogging, or when using Global Illumination or faked GI, your render may be overly tinted. To rectify this, take the render into Photoshop and adjust the levels a little to equal them out, or apply Auto Levels and blend it with the original (using Fade Auto Levels after applying) so that some of the original effect and contrast shines through. Alternatively, make a clone of the layer, apply Auto Levels to it and adjust its transparency.
  Specular Bloom
We can add specular bloom as a post effect, again completely automated by using Actions, by duplicating the layer, changing its blending type to Screen, clamping off the levels of the layer copy so that we have a harsh transition from dark to light (with very little light visible) and blurring the result. Depending on the result, you may want to drop the opacity of the layer a little.
  16bit Image Handling
We have been able to render out 16 bit images for a time now, but only now have we been able to play around with them properly in Photoshop. Due to the high amount of colours available we can get nicer levels and less banding artefacts when performing colour adjustments for print-resolution images.
Enlarge Screenshot Depth of Field Effects
Create nice Depth of Field effects on your renders by duplicating the original image several times and performing varying degrees of Gaussian Blur on each layer, ranging from, say, 30 to 1, halving each Gaussian Blur setting (e.g. 30, 15, 7.5 etc). Finally, add a gradient layer mask to each blurred layer so that the less blurred layers are more opaque in the “foreground”. The best way to arrange these layers would be the most blurred at the bottom of the layer set and the least at the top so we have an overlaid effect which also gives us more control when designing the layer masks. Instead of using Gaussian Blur, try using the new Lens Blur feature!
  Fish-Eye Lens Effects
Open up the image you want to distort, create a duplicate and create a white to black Radial Gradient from the center (white) to beyond the edges of the canvas (black) so that the edge colours are a mid to dark grey. Save this duplicate as a .psd file, and apply the Distort > Glass filter. Load in the saved psd file into this filter to distort the image and, ensuring the scaling is on 100%, crank up the Distortion. You may need to apply the filter another couple of times depending on the size of your image!

Initially published: Computer Arts magazine, Issue 94, March 2004.

Copyright © Pete Draper, March 2004. Reproduction without permission prohibited.