Rendering category

When you render - you do not see the current state of rendering... in best cases you see last render time and current part-number...

But what about current render state in percent (for image or animation)? Or current render time for current frame. Or estimated time for frame / animation?

Look at this screen - wow!
http://images.gamedev.net/features/reviews/modo202/5-shot-glass-render.jpg

Caustics are the only real way to make a shadow of a transparent object look more photo-realistic. This would be a great advancement for people working with transparent materials.

It would be really great to be able to see particles/halo textures in raytraced reflections & through transparent materials.

The idea is to be able to pause a render and then to be able to resume the render later, either during the current Blending session or at some future time after having exited and restarted Blender.

The render state could be kept in an external file (e.g. example.blend.renderstate) while render image could be maintained in its own subdirectory (much like what was done with particle caching) so renders could be continued not only on one system but multiple systems if necessary (assuming in this example the same version of Blender is being used on all systems involved).

That last part could also offer potential benefits for advanced renderfarm applications. One example being where a custom (i.e. Blender-specific) renderfarm with potentially multiple different systems (render nodes) could, upon finding calculating an opportunity (either predetermined via benchmarks before production time rendering or through realtime performance checks) for faster rendering on a newly available node with faster/higher computational capacity (e.g. multicore system/more RAM), terminate the slower render node (while not losing its progress by virtue of a rendestate) and then advance the job (blend file) to the better node with renderstate information so as to resume the render.

Just some thoughts :-)


Thanks for considering,

RobertT

Currently whatever renderer you use, Blender becomes unresponsive until rendering is complete.

Cinema 4D is the closest (interface wise) to Blender (total customization).

When one renders in C4D, i can still continue to work on the project, and the scene simply renders in the background (with the penalty of render speed) but overall productivity increases 10fold as I never (nearly never) stop working on the project.

The developers of Freestyle , a non-photorealistic renderer, have suggested that their renderer be integrated into Blender. Showing support for its implementation, here is one user's request to accompany the request of Freestyle's developers.

http://freestyle.sourceforge.net/

A loadable saveable render preset so you can load light positions, render settings etc quickly. Many of us use a lot of the same rig like the three point rig etc many times again and again. it'd be nice if we can simply load a preset and have everything setup and ready for render in a single click. It sure beats trying to figure out what setting I used before.

Perhaps something like that's available in vue where you'd just load your sky/clouds and select either draftm best or ultra quality render. Then hit render and have it look good all the time.

There needs to be an option for the Emit button to enable the material to not only be brighter but to light up things around it (similar to yafray). Also textured objects could give off the color of the texture.

Blender is in need of Render-time (renderer integrated) MicroPoly or Subpixel Displacement. Seeing that 2.46 will support 32-bit floating point maps for use with the Displace Modifier, it is almost essential that Render-time MicroPoly or SubPixel Displacement be an option with no dependency on the Displace Modifier. Look into Mental Ray's or Vray's* implementation.

Currently we can import nodes from a .blend file provided they have been grouped before saving.

Groups do not support Render Layers when attempting to select all nodes and creating a Group.

There have been times it would have been very helpful to be able to import entire render settings and complete node setups, especially where animations and settings (such as FFMPEG, OSA, render size, etc). can be concerned over multiple blends after a project became quite involved and altered over various stages of production.

Perhaps the upcoming architecture of Blender will enable such data being importable in future .blend files.


Thank you for considering this idea,

RobertT

It would be nice, if Blender would take an image of the 3D-Camera and put it in the background of the render window. So you could see the whole scene. And piece by piece the final render appears over the fast rendered background. It takes Blender ~0.01 secs to capture the image and set it as the background, so rendering wouldn't slow down, but you can see the whole Picture before its rendered. It's not very helpfull for animations, more for stills.

Currently it is possible to get results comparable to global illumination with the internal renderer but they require a lot of tweaking such as setting up a light dome and/or to apply radiosity. Until now using an external renderer such as yafray or kerkythea is a workable alternative. However, I still prefer the internal renderer as this is more tightly integrated. For example, cancelling the render process works without a problem with the internal renderer, with yafray this seems more difficult. Other problems are the difference between the light settings needed for yafray and needed for the internal renderer. That makes it difficult to create a quick preview using the internal renderer and then use the external renderer for the final picture. Therefore I propose to either adapt the internal renderer to have features comparable to GI, or to improve the integration of an external renderer.

Renderman is a well documented API with several, very good GPL as well as commercial standalone rendering systems available. It would be nice if Blender could talk with Renderman more eloquently complete with RM shader support, perhaps a "Renderman" tab could appear in the materials editor when a Renderman engine is selected.

A lot of Blender->RM Shader support has already been spearheaded with RIBmosaic. I'd like to see RM more tightly integrated along side Blender Render and Yafray.

The more options the better, and the fastest way to create those options would be Renderman integration.

Add a setting to the render section in blender that auto saves each fully rendered image automaticly in a set directory. It's a real nice way of keeping track of your progress, especially in combination with the "stamp" function.

Basicly the same 'save' function that is executed when you animate, with the exception that it should see what images with a matching name/type are already present in that directory. And if there are any, then add 1 to the highest nr found, and as with an animation call it for instance image_0002.jpg

I think it's also important it only saves when a render is completed, not when a render is cancelled (btw, it would be real helpfull if a distinction between those two could be made in the python api as well).

Here is an example python script that already does it as a scriptlink, which I've found extremely usefull: http://www.alienhelpdesk.com/python_scripts/autosave_render

It will be very useful if the rendering window can:

Has many rendered images saved in memory until we close it. So you can compare one with another to decide which is the best. And you can set the total numbers allowed so it won't consume much memory if you do large resolution rendering.

Has ability to let you put some notes on it. You maybe want to put your radiosity settings, MBlur value, light positions ... everthing important to generate that quality of rendered image.

Auto generated notes about the rendered images, like time needed to render, which renderer you use (Blender internal, Yafray, or maybe external), which camera used, antialising information, etc. All the items to show in rendering window can be chosen in Preference menu.

Has some photo editing features. No, I don't mean Adobe Photoshop - that is too feature rich. Just some simple brightness/contrast/gamma adjustment, blur, and color balance. You know you often do brightness correction after rendering. But sometime the image itself doesn't contain enought color range, if you adjust the brightness too far it will become dull. So you can test adjust in rendering window to decide to use it or reconfigure the lamps.

You can save the rendered images directly from rendering window, don't have to use F3 (of course you still can use F3 if you want). And you can make it autosave using an autoincrement filename to a folder you specify.

I've been doing some benchmark tests using SGIs (using
test.blend from www.eofw.org/bench), during which I got
to wondering about how Blender renders the sub-regions
of an image with multiple threads.

Assuming the use of N threads, once there is less than N
areas remaining (call it K) then some threads go unused,
so the tail end of render process is not as fast. Worst
case is if the final region happens to be a complex one
in the image: only one thread is running and it takes
much longer than normal. This wouldn't matter if every
subregion was equally complex, but in real images this is
never the case.

In other words, if there are N threads, the parallelism
drops off as soon as there are N-1 areas left to render.
Unless the overhead kills it, surely it would be better
once K < N to halve the width/height of the remaning K
areas, which would mean being able to use N threads again,
ie. maximum speed. Depending on the resolution of the
image, this could be done once or twice and should speed
up the rendering of the final N-1 pieces quite a lot.

Example: 8 threads (very common these days with the latest
dual/quad-core CPUs). Image split into the default 4 x 4
pieces. When 7 pieces remain, halve the width/height of
the peices, so thus 28 remain, and 8 threads can be used
again. As before, when only 7 pieces of this smaller size
remain, the efficiency will slide, but the final result
will be quicker than without. If the image is large enough,
a further resolution-halving would still be effective. At

[....]

The developers of Freesytle , a non-photorealistic renderer, have suggested that their renderer be integrated into Blender. Showing support for its implementation, here is one user's request to accompany the request of Freestyle's developers.

Standard video size types for Blender in menus like this:

Cinema :(2k,4k,8k)
HD :(480,720,1080)(P,I)
PC ( Default, preview, Full )
PAL (16:9, Standard )
NTSC (10:10, 10:11, 16:9)
PANO ( NORMAL, EXTENDED )

Like with all major 3d animations, it would be nice to have an integrated solution for rendering across a local area network. Such as under render having a box called "NetRender" after which it would scan the LAN and distribute work.

issues/disadvantages of external solutions
- external application, in some cases complex setup
- lack on integration for post Node processing (as far as I know)
- lack of development
- bucket rendering (in some cases) requires second apps
- not portable between computers


Last might need some definition

portable means that I simply copy one directory (blender) and no installation is required (windows). with other packages, additional setup is required.

when brecht demonstrated the new AAO on the peach blog
(http://peach.blender.org/index.php/approximate-ambient-occlusion/)
he spoke of it being able of indirect, 1-bounce lighting

this feature (even if it's incredibly slow) would greatly benefit the internal rendering engine...