OctaneRender® 1.024 beta 2.48 TEST (win) [OBSOLETE]

[jamnique]

It never ceases to amaze me how much you can do in such a short time, being such a small team. Congratulations. I wish i could afford to buy a couple licenses more ;]

[abstrax]

PMC is double time slower then PT

PMC seams to become very heavy comparing to ver.before

PT literaly eat all graph.card memory i never seen mes.like on pic

Older octane file didnt work on this ver. i had problem with camera. File was loaded but i was not able to move with mouse, only with keyboard input, numeric input in cam.setings

PMC and PT are somehow slow...i have feeling that i downgrade my card a little bit but that is probably cos i expected optimization of rendering engines

counter, including time of rendering is working, but bumpy...it works in waves...like my comp is choking

min_display_samples what is this ?
keep on good work

Regarding PMC: Could you please compare previous render times with current ones? Has it changed much? Our tests haven't shown big differences in speed compared to beta 2.47. It would be great if you could send us a scene where there is a noticable speed difference between beta 2.47 and beta 2.48.

Regarding PMC and PT being slow: Does that apply to interactivity or to render performance (i.e. render time)?

Thanks,
Marcus

[abstrax]

memory display. Abstrax said that now the memory display shows the available memory ,
but that cant be true . with my GTX 285 i could easily render project up to 900mb of
the 1000 mb available. now octane tells me i only have 1033 mb of 1500 available.
that's almost 500mb missing somewhere .(no other application running )
that must be a bug .
Rico

Hmm, the way how we do things is: We sum up all memory we have allocated on the GPU and add the amount free memory we can query from the GPU driver. The sum is the "usable/available GPU memory". If that number is very small, there might be 2 reasons: We either don't account for all the memory we have allocated or there is more memory used by other things, like OpenGL and framebuffers.

I will have a look if there is more information we can query from the GPU driver to verify the numbers. On my machine they have been very reasonable.

Could you also please have a look at the CUDA device manager (File -> Preferences... -> CUDA Devices). If you choose the graphics card in the device info combo box, the usable memory displayed is the actual memory you can use for that card. If you use several cards together only the lowest amount of all cards can be used since the we need to load all data onto each GPU. That lowest amount is the one displayed in the viewport status bar.

I hope that makes sense. Cheers,
Marcus

[abstrax]

>The new refreshing way is slow.

Perhaps there should be a slider to set the screen refresh time? 1 - 100s

>indications that Octane is actually rendering.

How about my mini traffic light idea?

edit: added some pics. 1 issue cant reproduce atm -sorry

We may add an option to allow the user specify the upper limit of the refreshing time.

Maybe let me explain how refreshing works now and why we do it like that:

Everytime you make a change on the nodes (viewport navigation basically just modifies the current camera node) the image display gets updated as soon as possible, i.e. as soon as there is at least one result available. After that the refresh interval is set to 100ms and gets increased over time up to 5 seconds. Each time the interval has passed a new result gets displayed (if available) and then we wait until the next interval has passed.

The reason we do this, is to reduce the time spent on tonemapping. At the moment tonemapping is still cheap and fast, but we might add more stuff to it, which probably will slow it down (because it needs to do more work). And fast refreshes only make sense for a limited amount of time after a change, because after you have reached a certain amount of samples/pixel you don't see many differences between consecutive frames.

Regarding the traffic light: The sampels/second display shows you the current samples per second and goes down to 0 when rendering is finished or paused. -> You just need to look at the samples/second to see if it's running or not.

Cheers,
Marcus

[abstrax]

hhello,

after some time (4 or 5 minutes) one GPU is fails. i can swich beetween card (remove one, add one in pref), but same result. after some time, render continus with only one card.

cuda v4, driver 275.33, w7-64bits
(temp card no fails 74°, card fails only 66° so is not temp fault)

That sounds like a bug to me. Does that happen with any scene or only with specific scenes? Does the status bar display "(1 failed)"?

Thanks,
Marcus

[pixelrush]

>We may add an option to allow the user specify the upper limit of the refreshing time.

well it wouldnt hurt to do this would it? particularly for big renders or where you are happy to leave it running for a while.
you do tend to wonder if its working or not though hence the traffic light suggestion you dont like...I guess another idea is to write the refresh period there...'2sec updates' or such..

[abstrax]

Material ball pop-up is the biggest problem here too, i pick a material from the viewport, change its type (i.e. diffuse, specular, glossy, material mix) the material ball scene automatically comes to view. it doesn't pop up only when i select "portal" (which works fine by the way)

I have fixed that issue. As far as I can see, it occurs only if you have selected the material you change via the material picker, or am I missing another situation?

Cheers,
Marcus

[Alain]

@radiance
Thanks for your explanation. I understand now why some functions don't work anymore in this release

I really appreciate your work and I'm positive for the future of refractive software because you realize a new way of rendering. I'm sure because of a team like you "the rendering business" in the future will be completely different than today and you will reach more REAL artists because rendering is getting simple and fast like Photography.
The artists can concentrade on the artistic aspects of an image instead of struggling around with comlpex unnecessary technical computerproblems.

So keep on doing this great piece of software, I guess none of your competitors have a faster unbiased engine yet


Kind regards
Alain

[Zay]

I noticed frames per seconds has been removed. It was nice to have to check the speed. Any chance to have it back?

We are not constantly refreshing the screen anymore and with that the idea of FPS doesn't make much sense. It would be possible to calculate frames per second out of the samples/second, but why don't you use the samples per second display? Or am I missing something here?
Cheers,
Marcus

How would you tell how long it will take to render 2500 samples now? When I set up a scene I do it in 500x500 pixels to have a very fast preview response time.
The scene might say 5Ms/sec. For final render I set the image size to 2000x2000 and it's now say 8Ms/sec. How would you calculate how much time it now takes to hit 2500 frames? If you know that, could you please add - something - that tells how much time is left?

[Zay]

No subsambling, 2x2 subsambling and 4x4 subsambling doesn't seem to work. They are greyed out.