Just a quick animation in Blender. The oriental Red Dragon symbol, as used in Mah-Jong sets.
- Rotating view
- A close up shot
Video here: Youtube
A week ago I decided to tackle Blender (the open source rendering and animation package) again, since I didn’t get anywhere the last time I tried. It’s quite a steep learning curve, though I gather it’s much better with this latest version. I decided that the three things I’d most like to use it to illustrate were (1) monsters, (2) robots, and (3) alien landscapes. (Once a science fiction fan – always a science fiction fan). Just as you can never have too many monsters, you can never have too many tentacles. This was an attempt to make a suitably alien tentacle. Not brilliant, but I think it’s a good start. Deciding to animate it – just to see if I could – was challenging but very satisfying.
I’m an engineer and computer programmer, and when I look at an image, I’m always mentally translating it into a table of numbers. Because that’s all a displayed image is – just an array of red/green/blue intensity values at N pixel locations. I think that carries over into my attitude to post-editing and cgi. It doesn’t matter to me whether an image was created by extremely careful photography, rougher photography + touching up in a photo-editor, rendering software, or even someone painstakingly filling in the pixels one by one with MS Paint. All that I care about it is the final image.
This attitude is nowhere near universal. I’ve seen websites (including blogs I’ve visited because the author ‘liked’ my blog) with a proud declaration “All images on this site are un-modified and just as they came out of the camera”. Which is fine – it would be a boring world if we all thought the same – but makes an interesting contrast. To me it smacks of entering the Guiness Book of Records for hopping on one foot to the top of a mountain. It shows skill and determination, but no-one can tell from the triumphant photo at the summit just how you got there – and it’s much easier to walk.
Raining hard outside so I decided to snap a photo of a pot plant and composite it onto a pile of rocks I photographed yesterday (as if it was a giant plant). I can’t say I’m happy with the result, but extra practice can’t hurt. A bit tricky extracting the cactus from the background, and masking three layers of rocks to blur the foreground, but it helps me get more familiar with GIMP.
I spent a chunk of the day reading the GIMP manuals and various tutorials. This “dripping paint” (link) tutorial was very simple, and wrapping it around a sphere made an interesting composition. This is just a few steps with a graphics editor, no 3D render or raytracing.
Since you can never have too many monsters, this is a quick example of making a monster face out of a harmless cloudscape. It may prove useful to someone in a somewhat bizarre set of circumstances. I’m doing these steps with “PaintShop Pro v7” but any raster editor that handles layers such as “GIMP” (free) or “PhotoShop” (spendy) will do it.
(1) A sunrise sky. The mixture of colours is important, though not the exact colours. Work with a copy. You may want to chop out buildings etc but a tree or two can probably be left in.
(2) The next step is to duplicate the image as a new layer, then flip it upside down. Now we have two versions of the sky above each other.
(3) Change the “blend mode” for the upper layer from “normal” to “difference“. This subtracts the R,G,B values of each pixel from the top layer from the R,G,B values of the pixel in the layer below. Since the two layers are mirror images and difference mode is commutative, the result must be symmetrical about the mid-line. (Note: “multiply” mode will have a similar effect but it tends to darken things too much).
(4) Merge (flatten) the two layers down to a single layer and rotate 90°. This may differ in other editors, but in PaintShop Pro, rotating a multi-layer image only rotates the top layer which is not what we want. Try 90° left AND 90° right to see which one looks more monster-like. The human brain has dedicated processing hardware (like the graphics processor in a high-end video card) that identifies faces and animal shapes, so it’s surprisingly difficult NOT to end up with a face. At this point you may want to crop the image a bit, go back to step (2) and try a horizontal flip, or give up and pick a different source image.
(5) If, as usually seems to happen, you have a face then the next step is to create a selection over the eyes. This image happened to have square-ish eye sockets so I’ve just used a point to point selection. Once you have your selection(s), shrink them down. I shrunk these ones down by 8 pixels.
(6) Add a new layer to picture and draw into the selections with a contrasting eye colour. Here I’ve used bright yellow. Don’t worry about how garish it is, the next two steps will sort that out. You may want to add a darker pupil in the centre. Accuracy isn’t essential.
(7) Hide the background image so only the eyes show and get rid of the selection.
(8) Sharp edged eyes aren’t what we want so apply some significant blurring. I used a 10 pixel Gaussian blur, twice. You may want to touch up the pupil a bit.
(9) Redisplay the background image under the eyes and drag the transparency of the “eyes” layer down until the eyes are visible but not standing out too much. I didn’t get them quite right in this example which is rather annoying as I did when I ran through the steps the first time. If one eye came out better than the other, you may want to mirror it onto the poorer eye.
(10) Just so the image looks fractionally less symmetrical, I’ve rotated it by 2° and cropped back to a rectangle. There you have it, a monster made of cloud. Silly, but fun.
And another example: