[REQ] "Blow Up" rendering ... ;o)

[psor]

Heya Nik!

Since we have region rendering now, I would like to see "blow up"
implemented too! I dunno what another name for it, it's called
"blow up" in 3dsmax. What it does is, that like region rendering
you can specify an area that you would like to be rendered, but
rather like region rendering it "blows" the region "up" to the full
size of the resolution you've chosen in your setup.

This is freaking cool when you would like to render out details
from a nice shot ...

I hope you can do that. Right now it would be possible too. You
could choose another resolution and render a region, but I guess
that the "black frame" would increase the memory usage for
nothing. Just my two cents ... ;o))




take care
psor

[ViennaLinux]

does not make that much sense because we are simulatin camera lenses right? so everything on the side is distorted - okay not when you shooting 1000mm images but lets say F=40mm.
I cropped my avatar picture from a wide angle shot and all the glases are leaning to the outside. And hey --- Blender does support multiple cameras .. I always add a lot of cameras and if I need a close up i simple activate them with [Ctrl] + [0].

[psor]

Hmm, ... but does the resolution depend on the cam?! I mean, wouldn't
it be possible to take the region render code and pimp it so it will render
not just the region but the region at a chosen resolution? I would say it's
possible. When the discussion about region rendering was going on on
the forum Nik asked if we would need something "around" the region
or not. The most said yes, so it's easier to composite.

So why not delete the "surrounding" and change the resolution for the
region you've chosen? Maybe I'm missing something tho. ;o)



take care
psor

[pixie]

Imagine that you have the render preset of 800x600 and region 200x150 (with the same ratio so we don't need to worry with anything else). The fact that it would become as large as the preset means nothing lens wise, it just means that the preset would now be 3200x2400 with a 800x600. There would be no distortion. The thing is, is really up to indigo to do it? IMO it is not, it would be up to the exporters.

[dougal2]

the exporter could do it, but then you'd need to chop out the region from the surrounding blank area.
in all honesty, that's probably the easiest way to do it.

[FakeShamus]

I would like to see this implemented as a way to start doing really big final renders with indigo. crashes due to ram limitations really set a cap on image size. that could be avoided if you could render out chunks of the image, without distortion from moving the camera around, and then piece them together. you could make really high-res finals.

[psor]

"the exporter could do it, but then you'd need to chop out the region
from the surrounding blank area. in all honesty, that's probably the easiest
way to do it."

I see a problem there, as FakeShamus already said. Even if you could crop
the region later, while you render, the black area is a part of the framebuffer
and that means --> "RAM eater" ... which is a problem!



take care
psor

[OnoSendai]

Hi Psor,
Implementing a 'blow up' region could be possible.
Tell me, what would happen if the blow up region and the full image has a different ratio of dimensions?
E.g. what should happen if you try and blow up a 100x100 region to a 800x600 image?
I guess I could just render the region as 600x600 with some borders.

[Woodie]

In 3dsmax blowup selection has always the same ration like the final rendering. In our case you probably would use the largest matching.
Lets say you have set your rendering to 800x600 (4:3) but you selected area 200x100 (2:1 screen resolution) it should detect center of selected area and use width as your base or if but you selected area 200x200 (1:1 screen resolution) it should detect center of selected area and use height as your base.
It probably could also have option to render selected region as it is with up-conversion: if but you selected area 200x200 (1:1 screen resolution) it should detect center of selected area and use height as your base but render 600x600 image (maybe with black sides to achieve 800x600).

[Kram1032]

Hi Psor,
Tell me, what would happen if the blow up region and the full image has a different ratio of dimensions?

Well... That might be rather simple:

From the start res, you multiply the X and Y dimension, so that you get the pixel count.
Then, you use the image ratio of the cropped image (X/Y) and divide (if ratio > 1) or multiply (if ratio < 1) the ratio and then divide the result through the pixel count.
Now, you have two results. Round them and you have your X*Y-cords...

At least if I didn't have a thinking mistake, right now... Though, if that wasn't correct, the solution should be fairly close to that