Author Topic: Daily Builds 1.0 - 1.4  (Read 244805 times)

2016-05-06, 16:28:27
Reply #1440

mitviz

  • Active Users
  • **
  • Posts: 482
  • Architectural/Product Modeling and Rendering
    • View Profile
    • MITVIZ
yeah i know, i wish i had a solution for it though, its happening when i use any version of corona now, such a shame i wanted to use it today but didn't want to risk it yet, guess i will try disabling fth again some other day and work that way but idk based on what i read if that's a good idea and it only occurs with windows 7 correct?

Hmm first, I would recommend checking your ram for errors, I hadn't got such serious errors and max file corruptions or fth build up or any other crap like that since 2012 when I bought my good old 3930k with 64 gigs (8x8gb) of ram with one of those 8 modules being faulty, which, being such, produced all kinds of weird stuff when ram consumption went past some threshold - 48 gigs or somewhere round that - maybe even less but I don't remember that amount precisely anymore. I didn't know what it was that ram module's fault as the thought what branded\radiatored\kitted and overall quite press praised and expensive ram can be faulty does not come thirst at least not to me and not back then. So until the day it's miserable existence finally ended and it disappeared from both bios and windows shown memory completely, I was getting file corruptions, render crashes (I used vray at that time) and some general system instabilities and was trying all kinds of stuff to fix that - even considered disabling overclocking for a moment (what a stupid thing to do)) which, of course, didn't help and generally battled with various imaginary windows problems, imaginary malware and viruses software conflicts etc. yet in the end it turned out it was that goddamned ram module's fault all along. Second thing to try would be disabling antiviruses, realtime tweakers\hdd defragmenters\ram optimizers\drive backup softwares\etc (if you use any) as those could possibly cause such things too. And third would be checking hdd for errors bad sectors etc. - hell I even got some faulty sata cable some long time ago which caused some similar crap )) Well as you can see there's a plenty of causes for such behavior and quite a plenty of things to check but, imho, ram being the main culprit, so good luck finding your specific one.

corona guys actually replied to my ticket and i got some of these are advice, will be a hunt to find the culprit but i really hope it doesnt happen again over the next few days or i will go crazy, too much going on to check all these things but i will!!! and thanks for all the replies!
Mitviz
Professional Modeling and Rendering Solutions
Intel core i7-5960x cpu @3 GHz, GeForce GTX 970

2016-05-06, 16:36:19
Reply #1441

Noah45

  • Active Users
  • **
  • Posts: 437
  • press *1
    • View Profile
Simple, but effective way not to lose data. Preferences>increment on save. Hell, I turn off auto backup, and get to the habit of; "I'd hate to lose that thing i just did", CRTL/ SAVE+
Retail Illustrator  (for ever' 80's )
3DMax 2020/Corona Version: 6DB

2016-05-06, 17:54:45
Reply #1442

Juraj

  • Active Users
  • **
  • Posts: 4743
    • View Profile
    • studio website
I have some trouble understanding remaining render time estimation. It seems that with adaptivity enabled you can't trust the information displayed, I have renders where it will show 2 hrs remaining rendertime for the first 5 passes, and then starts growing until it displays something around 20+ hours. If I disable adaptivity, the remaining estimation will behave like expected, it just counts down as time passes. The final result is that adaptivity adds a lot to the rendertime and the final rendertime increases by a factor of 10 on a big image here vs. a render without adaptivity. The final image may look slightly better but a factor of 10 for slightly less noise is in no way right. Plus, you can't rely on the estimated remaining time since it is misleading.
Given that adaptivity affects render times that much it should probably be moved from devel/debug section and not be enabled by default.

Seems odd.. the adaptivity really took 10times longer for same noise ratio (or passes) ? Or is that only based on estimate ?

Please follow my new Instagram for latest projects, tips&tricks, short video tutorials and free models
Behance  Probably best updated portfolio of my work
lysfaere.com Please check the new stuff!

2016-05-06, 18:04:14
Reply #1443

pokoy

  • Active Users
  • **
  • Posts: 1850
    • View Profile
I have some trouble understanding remaining render time estimation. It seems that with adaptivity enabled you can't trust the information displayed, I have renders where it will show 2 hrs remaining rendertime for the first 5 passes, and then starts growing until it displays something around 20+ hours. If I disable adaptivity, the remaining estimation will behave like expected, it just counts down as time passes. The final result is that adaptivity adds a lot to the rendertime and the final rendertime increases by a factor of 10 on a big image here vs. a render without adaptivity. The final image may look slightly better but a factor of 10 for slightly less noise is in no way right. Plus, you can't rely on the estimated remaining time since it is misleading.
Given that adaptivity affects render times that much it should probably be moved from devel/debug section and not be enabled by default.

Seems odd.. the adaptivity really took 10times longer for same noise ratio (or passes) ? Or is that only based on estimate ?
No, this is what actually happens. The bad thing is that you may trust the rendertime estimate number, but beginning with pass 5 (probably due to adaptivity being rebalanced every 5 passes) it'll begin to climb up, and will keep rendering for 20 hrs.
My description of the issue is a bit convoluted but in simple words: render without adaptivity takes 2 hrs, render with adaptivity takes 20 hrs. Happens with daily 24-Apr-2016 and 1.4 RC4.

I can't share the scene, I know this is very unfortunate for you the devs. But I bet this can be observed on other scenes.

2016-05-06, 18:25:11
Reply #1444

sevecek

  • Former Corona Team Member
  • Active Users
  • **
  • Posts: 197
    • View Profile
But I bet this can be observed on other scenes.

To some extent. Adaptivity makes renderer to focus on problematic parts of the image that take longer to render than, say, directly visible HDRI map. But unless your image is 90% HDRI map, this probably isn't the cause of the problem.

Could you try to render CInfo_SamplingFocus render element? That may be a clue to what's going on.

2016-05-06, 18:33:16
Reply #1445

pokoy

  • Active Users
  • **
  • Posts: 1850
    • View Profile
But I bet this can be observed on other scenes.

To some extent. Adaptivity makes renderer to focus on problematic parts of the image that take longer to render than, say, directly visible HDRI map. But unless your image is 90% HDRI map, this probably isn't the cause of the problem.

Could you try to render CInfo_SamplingFocus render element? That may be a clue to what's going on.
Thanks for the advice - will do on monday!

2016-05-06, 18:33:42
Reply #1446

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8779
  • Let's move this topic, shall we?
    • View Profile
    • My Models
No, this is what actually happens. The bad thing is that you may trust the rendertime estimate number, but beginning with pass 5 (probably due to adaptivity being rebalanced every 5 passes) it'll begin to climb up, and will keep rendering for 20 hrs.
My description of the issue is a bit convoluted but in simple words: render without adaptivity takes 2 hrs, render with adaptivity takes 20 hrs. Happens with daily 24-Apr-2016 and 1.4 RC4.

Rendering can take 999+ hours with adaptivity or without. You can't directly compare passes to passes when adaptivity is involved anymore. If you used to trigger end of rendering process with pass limit, why not o try to use time limit instead?

Maybe pass limit could be removed from render settings or hidden, to not bring confusion and complaints amongst users?
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2016-05-06, 18:38:50
Reply #1447

pokoy

  • Active Users
  • **
  • Posts: 1850
    • View Profile
Sure, it's just that no one would expect the rendertime to rise by a factor of 10 by simply enabling adaptivity for the same image. The thing is that the image is rather dull, it really doesn't have that much contrast except for a small area where metals reflect the sun (which could be why this happens).

2016-05-06, 18:57:28
Reply #1448

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8779
  • Let's move this topic, shall we?
    • View Profile
    • My Models
You can't say that rendertime has grown up by the factor of 10, unless both images in the end has same SNR. But then it should be really something very wrong and it should be reported as a bug immediately.
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2016-05-06, 19:05:18
Reply #1449

pokoy

  • Active Users
  • **
  • Posts: 1850
    • View Profile
You can't say that rendertime has grown up by the factor of 10, unless both images in the end has same SNR. But then it should be really something very wrong and it should be reported as a bug immediately.
Can't compare SNR since the one without adaptivity doesn't display the noise percentage (good that you requested this).
As for the bug report - I would do more thorough testing if I had the time, unfortunately I can only spend half an hour max each day on issues like this. By posting here I also hoped that others may feel encouraged to do some tests. I know I'm not very helpful with this but time is the limiting factor for me.

2016-05-06, 19:22:20
Reply #1450

emanfrin

  • Active Users
  • **
  • Posts: 10
    • View Profile
Denoise does not affect   "Shadow Catcher " alpha channel   ?   

beauty render is DENOISE ok
alpha matte render   (Shadow Catcher )  composite mode no denoise, and very noise alpha shadowns .

or it is made in a different way?


and congratulations corona 1.4 this amazing  \o/ \o/ \o/

2016-05-06, 21:42:45
Reply #1451

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 9048
  • Turning coffee to features since 2009
    • View Profile
I have some trouble understanding remaining render time estimation. It seems that with adaptivity enabled you can't trust the information displayed, I have renders where it will show 2 hrs remaining rendertime for the first 5 passes, and then starts growing until it displays something around 20+ hours. If I disable adaptivity, the remaining estimation will behave like expected, it just counts down as time passes. The final result is that adaptivity adds a lot to the rendertime and the final rendertime increases by a factor of 10 on a big image here vs. a render without adaptivity. The final image may look slightly better but a factor of 10 for slightly less noise is in no way right. Plus, you can't rely on the estimated remaining time since it is misleading.
Given that adaptivity affects render times that much it should probably be moved from devel/debug section and not be enabled by default.
do NOT disable adaptivity because it makes your passes-limited render slower. Do noise limited render instead. Of couse because the adaptivity forces the renderer to focus on harder areas the render time to render X passes get longer, but the noise clears faster overall
Rendering is magic.How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2016-05-07, 10:44:10
Reply #1452

vansan

  • Active Users
  • **
  • Posts: 57
    • View Profile
Ondra, hello.
I'm getting a little different results with wood material bump mapping set on 0,001 value.
I attach images below.

2016-05-07, 11:05:27
Reply #1453

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12711
  • Marcin
    • View Profile
For me the difference is so subtle that it could be just noise pattern. It would be great if you could re-render at higher resolution (could be a crop).
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2016-05-07, 12:07:17
Reply #1454

vansan

  • Active Users
  • **
  • Posts: 57
    • View Profile
I guess that you are right.
Texture is looking a bit sharper and has better contrast in 1.3 comparing to 1.4 (both textures have 1.0 blurring)

updated: Found the way out - I set blur = 0,01 for all diffuse and reflection textures to get the same sharp look.
« Last Edit: 2016-05-07, 12:29:58 by vansan »