Maxwell Render | Official public forum

I'm being asked to do a 12,000 pixel render (which I know is over the top). I can't render this here and I don't think most render farms can handle that either... I can't even start the render to predict a cost.

How do I go about splitting it up and rendering out 3 images? Do I just use 3 cameras (for example) and then use Shift Lens to match them up side by side...?

Hoping someone can give me a quick reply as he want's it done within a day or so.

Thanks.

unbiased rendering will show different results, if not the whole scene is one rendered frame..

but i never tried it so i don't know how harsh the differences are, it will for sure depend on the scene and it's lighting.

you could use lens shift with a good overlap,.. so you get some space to improve the transition in post... and make sure vignetting is off

Testing it now.

How come Maxwell does not render out the camera view point after I have changed the Shift Lens value? It doesn't render the yellow viewpoint area.

maybe it's because you can't shift more than 100%?
(which also means you cannot render out a small rectangle in one corner)

I remember I once did experimented with something similar.
I used 'render region' and stitched the parts afterwards in PS - there were no visible signs of stitching.
But SL must be EXACTLY the same for all parts.
And also I used 1.7; don't know if V2 will handle it the same way.

kami wrote:maybe it's because you can't shift more than 100%?
(which also means you cannot render out a small rectangle in one corner)

Well I only shifted it 50%.

m-Que wrote:I remember I once did experimented with something similar.
I used 'render region' and stitched the parts afterwards in PS - there were no visible signs of stitching.
But SL must be EXACTLY the same for all parts.
And also I used 1.7; don't know if V2 will handle it the same way.

Oh yeah... I've only just realised you can render out Render Region. I thought that was just used within Studio.

I've had to do renders such as these before for clients (i just did a 10,500px render last night...it was being used in a brocure WTF?!?!) My advice would be to not render full res. I would so something like 8,000 then scale it up. The client will most definitely will not be able to see a difference between that and rendering full res. Considering how many pixels you have to begin with, you can scale quite a bit without it becoming blurry. Unless there is some weird reason why it ABSOLUTELY has to be done at full res, i would save yourself some headaches and render smaller

itsallgoode9 wrote:I've had to do renders such as these before for clients (i just did a 10,500px render last night...it was being used in a brocure WTF?!?!) My advice would be to not render full res. I would so something like 8,000 then scale it up. The client will most definitely will not be able to see a difference between that and rendering full res. Considering how many pixels you have to begin with, you can scale quite a bit without it becoming blurry. Unless there is some weird reason why it ABSOLUTELY has to be done at full res, i would save yourself some headaches and render smaller

Yeah I'm doing that too using Alien Skin Blow Up. Hope he doesn't keep watch on this forum.

Another thought regarding stitching the images together. You might also want to set the initial CPU seed value to the same number for each of the renders as well. That should keep the noise the same on each of the images. Not sure if not doing that would be noticeable but, as that sort of rendering is something you only want to do once you'd better take every precaution you can.

-Brodie

brodie_geers wrote:Another thought regarding stitching the images together. You might also want to set the initial CPU seed value to the same number for each of the renders as well. That should keep the noise the same on each of the images. Not sure if not doing that would be noticeable but, as that sort of rendering is something you only want to do once you'd better take every precaution you can.

-Brodie

It seems to work ok just by reaching the same sampling level, but how do you do that anyway?

I'm not sure if you can do it from studio or plugins. But from Maxwell Render you set the CPU ID



-Brodie

You can add a command line function:

cpuid:1

But I don't think you will need it.

Ah right ok, didn't know about that!

its a small world

thx for the affront and greetings from the dickhead
I hope you found a way to render your 12k pixel image yourself