Nevermore
06-17-2006, 01:06 PM
Just in case anyone is interested. I posted this at another site but thought kindreds might have some interest too. Go straight to Giordan's The Art of Photoshop if you have it for a professional take on this, he is brilliant but if you don't have any access, here is a brief summary:
A displacement map ("DPM") lets you mould one image to another. Say you have a pumpkin (a neat textured surface) and you want to superimpose a witch on it as if she was painted on. You could just overlay the witch and lower the opacity but she will, of course, be quite faint. If you make a DPM out of the pumpkin, you are turning its surface into a guide that changes (displaces) the pixels in the witch by slightly offsetting them to match the DPM information when you run the DPM over the witch using the distort filter. So your witch really looks like she is painted on the pumpkin surface because you used that very surface to distort your witch. What is tricky is guessing how much to offset the distort filter. The grand finale is playing with the filtered witch by using blend modes and masking (that lets bits of the real pumpkin show through). Or whatever.
1. You need a target object. Say a pumpkin.
2. You need an image you want moulded onto the target object. Say a witch.
3. You need a DPM. Make a duplicate of your target object (pumpkin). Turn it to greyscale and save it somewhere handy as a .psd file. This is your displacement map. Giordan blurs it slightly using a Guassian blur of about 4 before saving it as DPM. I do too just because I haven't had time yet to figure out if it is really necessary or what would happen if you used a different blur or a higher Guassian blur. If anyone gets through those options, PLEASE let me know your results.
4. Put your image (witch) on the target object (pumpkin) as a new layer.
5. Apply the DPM to the image layer. This is hell on wheels because you are doing it "blind". When you chose Filter-Distort-Displace the first thing that comes up is an option box asking you how much you want to offset the image (witch) pixels. Who knows? I do it by 5s ie I try 5 and then 10 and then 15 until I get what I want. I use the same number for both the vertical and the horizontal offsets, again because I haven't had time to play with different values. Again, I beg anyone who has to let me hear about it.
6. Once you pick those options, you get asked where is the DPM you want to use and of course, you must browse to where you saved the DPM you just made out of your target object. Viola! (Tip: sometimes the result can be subtle but it is very much there. To check it out, click undo and redo several times and you will see your witch straightened, displaced, straightened, displaced. I gotta get a life!)
7. You will then probably spend a lot of time undoing the filter and running it again with different offsets values.
8. Of course you can mess with the other options in the displacement input table but, you guessed it, I haven't had time. You can also get pretty funky results by displacing and then using different blend modes on the displaced image. I find the whole process mega entertaining and time consuming which is why I haven't figured out enough of the variations yet.
Please post any interesting work or observations if you mess around with this.
A displacement map ("DPM") lets you mould one image to another. Say you have a pumpkin (a neat textured surface) and you want to superimpose a witch on it as if she was painted on. You could just overlay the witch and lower the opacity but she will, of course, be quite faint. If you make a DPM out of the pumpkin, you are turning its surface into a guide that changes (displaces) the pixels in the witch by slightly offsetting them to match the DPM information when you run the DPM over the witch using the distort filter. So your witch really looks like she is painted on the pumpkin surface because you used that very surface to distort your witch. What is tricky is guessing how much to offset the distort filter. The grand finale is playing with the filtered witch by using blend modes and masking (that lets bits of the real pumpkin show through). Or whatever.
1. You need a target object. Say a pumpkin.
2. You need an image you want moulded onto the target object. Say a witch.
3. You need a DPM. Make a duplicate of your target object (pumpkin). Turn it to greyscale and save it somewhere handy as a .psd file. This is your displacement map. Giordan blurs it slightly using a Guassian blur of about 4 before saving it as DPM. I do too just because I haven't had time yet to figure out if it is really necessary or what would happen if you used a different blur or a higher Guassian blur. If anyone gets through those options, PLEASE let me know your results.
4. Put your image (witch) on the target object (pumpkin) as a new layer.
5. Apply the DPM to the image layer. This is hell on wheels because you are doing it "blind". When you chose Filter-Distort-Displace the first thing that comes up is an option box asking you how much you want to offset the image (witch) pixels. Who knows? I do it by 5s ie I try 5 and then 10 and then 15 until I get what I want. I use the same number for both the vertical and the horizontal offsets, again because I haven't had time to play with different values. Again, I beg anyone who has to let me hear about it.
6. Once you pick those options, you get asked where is the DPM you want to use and of course, you must browse to where you saved the DPM you just made out of your target object. Viola! (Tip: sometimes the result can be subtle but it is very much there. To check it out, click undo and redo several times and you will see your witch straightened, displaced, straightened, displaced. I gotta get a life!)
7. You will then probably spend a lot of time undoing the filter and running it again with different offsets values.
8. Of course you can mess with the other options in the displacement input table but, you guessed it, I haven't had time. You can also get pretty funky results by displacing and then using different blend modes on the displaced image. I find the whole process mega entertaining and time consuming which is why I haven't figured out enough of the variations yet.
Please post any interesting work or observations if you mess around with this.