[REQ] Render Cropped Section

[Zom-B]

Hmmm... region rendering...

I have to admit, that reaching the goal to concentrate rendering Power on certain areas could be solved much convenient and productive (yeah, and more complicated to code for sure!)

Having the option to weight rendering priority to some materials used in the Scene on-the-fly (using GUI sliders) would be a very interesting...

After rendering an Interior for 10h you could concentrate Indigos Raytracing to focus on the SSS Material you used for some Plastic toys, or the Glossy Transparent Vase on the table...

I hope you get the point, I'm not telling this region rendering approach you are planing is bad, but its quite not that easy and flexible to handle than you guys think...

[SimonLarsen]

ZomB

But its a hell lot more diffucult to concentrate on some selected objects/materials.
Should indigo then calculate which "rays" would hit the selected materials and concentrate on them, and then what about caustics, mirrors and stuff.

Im really no programmer and doesn't know very much about raytracing and stuff, but this seems pretty damn complicated.

[suvakas]

I can either make the entire image size shrink to fit the render region, or I could draw the render region inside a large black image.

If you want to make it user friendly, then please let's do this like the rest of the world - render region inside a large black image. Otherwize the comp over final image would be hard to do.

[Marcofly]

Otherwize the comp over final image would be hard to do.

Suv, there's no difference! Assume you have a scene with a glass that needs more cooking time than the environment: you need to create a mask that has the shape of the glass, and it's not a square! so, who cares if the background outside the render region is black or pink? (i hope i explained it in a clear way..)

I agree with ZomB: i see the advantage of a render region only to make a test of 'difficult' materials, but i don't think that it will so useful as you can think..

just my 2 cents..

[Kram1032]

Would it also be possible to do it in a large transparent image?
That way, one could fill it afterwards, if wanted and it definitely has the correct position

@Macrofly: I think, suv meant, it's easier to align the cut-out image with the complete render, if you have the black frame. If it's reduced to the minimum square size, it'll be hard to place the images, to correctly match the scene

[suvakas]

@Macrofly: I think, suv meant, it's easier to align the cut-out image with the complete render, if you have the black frame. If it's reduced to the minimum square size, it'll be hard to place the images, to correctly match the scene

Exactly
Thank you Kram.

[Kram1032]

See?
I read just fine

[Marcofly]

ok, now I understand what you meant!!
The Kram-Webster-Dictionary works good..

[Kram1032]

Thanks

[noyb]

Hi All,
I'm working on region rendering right now, it's not too hard, although there are quite a few subtle complications regarding Reinhard tonemapping, resume rendering etc..

What I need to know, is what kind of workflow would be useful for the artists out in IndigoLand.

I can either make the entire image size shrink to fit the render region, or I could draw the render region inside a large black image.

How are people expecting to merge a full render, combined with a render-region render? Photoshop? Or is some kind of merging tool desired?

I was thinking of something simple like:

Code: Select all

<crop>
     <1>x y</1>
     <2>x y</2>
</crop>

Where you give 2 points on an image of a given width and height.

As for the tonemapping, I'm happy with camera tonemapping
I was thinking of just compositing the images in photoshop, but a tool may streamline the process a bit.
For compositing, I think it might be easier if the region is not placed within a black image - although it would be easier to position - perhaps an alpha channel value in the non-rendered portion to allow for easy masking.

Just thoughts - I'm sure anything would be great, it usually is

[conartist6]

Excuse me, because my concept of what i'm talking about is limited and i think in logic instead of computer code, but wouldn't bidirectional pathtracing make concentrating rendering power on a portion of the image fairly easy? Especially if we assume that the rest of the image has essentially resolved, rays could simply be shot back toward the light source/rest of the scene. from the geometry in question. There's probably a reason this can't be done if nobody else has mentioned it yet, so maybe somebody can enlighten me. Obviously, I think that this feature would be very useful as I have had renderings which could use the feature to significantly improve speed. Thanks.

[dougal2]

this feature was implemented a release or two ago.

[conartist6]

Oh, thanks. I'm rather new to Indigo. My only other question then is, if its IN the program, how the heck do I do it? I'm using Blendigo to export IGSs.

[dougal2]

I don't know. Read the blender forum, and blendigo release threads.