Charly 3D Tutors

  In this video, I'm showing a render-test of the new Redshift 3.0.43 on my dog project. The increase in speed on this scene: in one test - x1.7, and in the other x2.8 !!! And this is awesome! Since with the last test with the scorpion girl - I was a little upset, I was doing different tests all weekend - and threw them here on the forum - https://www.redshift3d.com/forums/viewthread/33548/#226035   During the tests, I came to the conclusion that RTX ON still needs to be improved in Redshift and not all tests are successful. We are already discussing these points with NVidia and the Redshift Team. I think that in subsequent releases, the situation will improve on those scenes that did not show acceleration ... Perhaps the reason for the slowdown in RTX ON in previous tests was the choice of the Hilbert method + increasing the bucket size to 256 ... I don't know. We need to check ... But here I left the default method - Spiral and bucket = 128 and it got a strong acceleration...

  In this video, I am showing a render-test of the new Redshift 3.0.41 on my recent project, which is already familiar to you from my previous tests. Since it has a new Hair Physical Material based on Disney research and I wanted to check how much slower it will render. Why was I already sure it was slower? Because many have already complained at the forum, but also since the material has become physically correct - naturally, it transmits more light and requires big miscalculations. But honestly, I didn't really get into all this .. I'm not a physicist or a mathematician. I am a creator with a creative mindset .. Therefore, for me - if only it renders beautifully and quickly, and what is there in physics and what processes are taking place is not very interesting =) Info about RTX SpeedUP - https://bit.ly/2R98DwF   The release of new NVIDIA Studio drivers also pushed me to this test. After all, Optix 7.2 was added to them, in which they promised to speed up the rendering of...

  I continue to record my tests on the GPU. This time I decided to test my recent project on 2x2080TI + NVLink. Shown a variant with CUDA and RTX. But there is clearly something wrong with RTX, since, logically, the scene should be rendered 2-4 times faster, because this happens on ordinary scenes without hair. But apparently, the rendering has the same problem as ChaosGroup with V-Ray. Hopefully, it will be fixed in future releases. I mentioned this bug on the forum - https://www.redshift3d.com/forums/viewthread/33548/ This video has no purpose to humiliate other renderings. I have already repeated this 100 times and will repeat it again. I'm just curious to check the performance of different renders in different situations - on my video cards. And yet - I have no purpose to compare renders. I do not like to do this, because I believe that everyone is good in their own way and everyone has their own advantages, and I never pay attention to the minuses. Therefore, I'm reco...

Redshift performance test in Houdini. The idea of this test came about after Andrey Flat showed a screenshot of this scene and I was impressed with the fact that Houdini @ Redshift3D very quickly turned 30 million instances * 4 million polygons = 120 trillion polys in the viewport. Yes, yes, I understand that there are only 4 million polygons in the viewport ... But then again, when I multiplied the instances so much in 3ds max - you all remember that I didn’t jump above 400 million polys. There was a crash on copies and instances ... And here - it behaves calmly at 120 trillion !!! View Andrey’s portfolio by clicking on the link - https://www.behance.net/flat4ers

Since literally all render developers add NVLink support, I decided to test it in Redshift 3.0.21. Because on this version - I noticed a strong jump in speed in Redshift Render. Many even talk about this on the forum =) The essence of the test is to fill the entire memory of one 2080Ti so that NVLink works and starts taking the missing memory from the second video card. Since Instance doesn't take up a lot of memory - I do a Copies. If you wish, you can enable Instance and render billions, but the essence of the test is not the point. I know about the Array Tool and when I tried it - Max crashed immediately, even on Object -Display as Box. So I carefully copied using Shift so that nothing would fly out. So far I decided to test without textures. The following tests will already be with them. And yet - I have no purpose to compare renders. I do not like to do this, because I believe that everyone is good in their own way and everyone has their own advantages, and I ...

Redshift test on the new Forrender.com servers PAYPAL - https://www.paypal.me/artofcharly WEBMONEY-Z861481193973 PATREON - https://www.patreon.com/charlytuts Hello everyone! With you Krivulya Andrew aka Charly and I record this video about the update of our renderfarm Forrender.com We have 3 new GPU servers. 4 * 980Ti, 4 * 1080 and 4 * 1080Ti. In this video, I conduct a render test of my scene on 980Ti and 1080Ti, as well as the standard Redshift Benchmark. In the comments write wishes, questions, topics of lessons that you would like to see in new releases, etc etc