I haven’t updated in several days! What’s happening with the perspective project you may be wondering? Let me explain what’s happening. So back several days ago, after the critique, I began playing around with the scenery again… which resulted in unusual renders. Here are some below.
… wow. But I see the problem. First of all, the image was not lined up correctly with the image itself. While I made sure that the settings in the image plane were the same in the projection node, I did not realize that if these two don’t mach, you obviously get a messed up result as you see in the two results above.
I decided to take a notch back. I’m not giving up just yet, no, NEVER giving up because this is what I wanted to do to begin with. So I thought I should find another image and give it a shot.
Here’s the new image I’ve been using since last week. This image is just easier to work with (I personally think.) So lets start with the basics. I first begin by drawing lines. Not just any line, a perspective line so I can tell where the horizon line is.
Great! Now, I want to note a couple interesting things I noticed about Japanese architecture. Unlike a grid-based approach to city planning, Japan takes a more organic approach to city planning. Meaning, that the city respects properties owned by people. Which results in a very complex system of roads and “blocks.” This makes home making incredibly weird since you want to use the maximum space possible while abiding all building laws. So, you end up building a house conforming to the roads, the given land space, and laws. This results in the unequal, shifty home design that breaks that clean grid look. Nothing in that image is a straight line. It’s somehow curved slightly.
So what does that have to do with anything? Basically this: determining perspective from squiggly lines from uneven building constructions and roads are hard. I had to really nail it but even with these perfect looking lines, the whole perspective is off. But whatever. It seems relatively ok, and I know I don’t have to go into so much depth.
Either way, I followed the same steps as above, I determined the sensor size, focal length, and finally the perspective on the camera. Then I began throwing some simple geometry down. And this is what I got.
After five attempts over the course of 3 or 4 days, I got this to work. And it was ridiculously easy. I wondered why it was so easy that I wondered what the trouble was after all. But ok, lets get back on track. Now we want to cast some textures on these simple geometries.
Next step: Render
All that trouble really paid off. The quality is a bit low, but that’s because the image quality to begin with wasn’t that high. We need to obviously fix a couple things, but overall, it works.
Lets try changing the angles a bit.
In the end we have limitations to this approach. You can see where I drew the arrows, the image is doubling. We also can see some significant blurring. This can probably be solved by using a larger image resolution at the cost of render times.
It’s been a very long post, but this is basically where I am now and what I’ve been hacking at since last week. Overall, it’s pretty easy BUT it gets hard after a while, I mean very tricky. Now I’ll have to edit the photograph so it won’t have any telephone poles and otherwise. I can add those in later as a separate layer.
Where next? Next? I’m planning to obviously get back to that painting somehow. I have no clue if it will work, but I know if I take it step by step it’ll work. Otherwise I’ll force it to work.
The other thing I was thinking of was extracting the textures from each face of the building for UV mapping. And I may go about doing that for this project. So next part: 1. Getting that painting working. 2. More photos to have fun with. 3. Trying a further defined approach (texture extraction from photographs.)