The Renderfarm

« first day (1877 days earlier) ← previous day next day → last day (2123 days later) »

12:01 AM

Hello

12:21 AM

man, the subtleties of building arrays totally confuse me. i managed to get one to wrap around itself literally about 20 million times.

now it's happy, because it doesn't have to use both a curve and an object offset. the nerve of me.

Duarte Farrajota Ramos

12:34 AM

It is all about object centers. If before starting all objects centers coincide in the scene, then you can expect predictable output. Applied scales also help.

kim holder

yep, the scale was applied and the curve had the same origin point.

but we understand each other now

Duarte Farrajota Ramos

12:51 AM

That looks an awful lot like this answer from a while ago

A: Object on curve wont stay flat!

If I understand correctly the deformation you are getting is cause by the spiral curve being naturally twisted itself. You can probably minimize this by changing the curve tilt per vertex (which would be very cumbersome), or overriding it's twist mode. In the Properties Window > Object Data (Cu...

kim holder

oh, interesting - i didn't know to check the curve's shape properties and try using Z-up. I'll have to bear that in mind.

i was a bit surprised i could do it without using a curve at all. but it is simpler.

but i too struggled with this for a while. i tried to get it to work with dupli-faces, and then with a few different sets of arrays before it worked.

gandalf3

1:31 AM

@GiantCowFilms I.. can't imagine that working very well

What even counts as a "mistake"

@Amir I haven't touched any of the physics code myself, but source/blender/physics has a suggestive name. Also source/blender/modifiers/intern/<whichever_physics_modifier_your_interested_in>‌ may not be a bad place to start

@Amir In my experience the best way to actually get a hold of one of them is to hang out in IRC and ask your question from time to time (it can take some experimenting with times of day, since everyone seems to be in different time zones). That seems to have a higher response rate than email at least ;)

2 hours later…

GiantCowFilms

3:34 AM

@gandalf3 It would be hard to get it work well, but I think there are somethings that are pretty easy to classify as mistakes. Start there, and refine. It won't be perfect, but it would help.

gandalf3

hm..

(unrelated): Speaking of questions which turn out to be new problems, I just found one:

Q: Object surface be cracked when move too far grid center

I have a problem but i dont know how to describe it. I have one object. It is look good if I put it in (0,0,0) xyz axis or about +- 100 unit location. But when i moved it to (1000,0,0) it became weird with some cracked on surface and it worse than that. I applied locrotsccale already, just subsur...

mesh

Considering how quickly it shows up (only 1000 units??) I'm surprised I've never hit it myself

(It's known)

GiantCowFilms

Close as bug?

Interesting though. Thanks for sharing.

1 hour later…

Duarte Farrajota Ramos

4:43 AM

Gotta learn how to make these seamless

VRM

4:55 AM

that's kinda cool

Duarte Farrajota Ramos

5:22 AM

thanks :)

batFINGER

Holy sperm

8 hours later…

Duarte Farrajota Ramos

1:10 PM

@batFINGER hahahah Cannot unsee

kim holder

2:01 PM

@gandalf3 most of the time i'm working at a scale of up to 10,000 units, and i've never run into it either.

1 hour later…

Amir

3:12 PM

Really appreciate @gandalf3. What was Blender Developer's IRC channel name?

batFINGER

#blendercoders

4 hours later…

gandalf3

7:02 PM

@kimholder It only happens when you are far away from the origin and have a small object

kim holder

yeah, i have human figures around, some of those are at least a couple thousand units away from the origin.

gandalf3

really, interesting

kim holder

i have little pumps and things like that, i think they'd fall within the right size range

gandalf3

indeed

kim holder

maybe i'm not quite far enough yet and i'll have to watch for it.

gandalf3

7:06 PM

I can see it at a 1000 units just by moving a normal Suzanne out there

I have to run now, but I'd be interested in maybe having a look at one of your scenes later, if you would be willing?

(just one placeholder object would be fine)

kim holder

sure, my pleasure.

gandalf3

later o/

kim holder

there's a quickie screenshot. i'll set up a better set of examples

3 hours later…

Duarte Farrajota Ramos

10:34 PM

Im testing here with a downscaled suzanne about .26 .16 .19 units at 100 000 units away from the scene center. No visual glitches on the mesh so far, but viewport navigation is starting to be an issue.

Rotating view is ok-ish, but object center dot glitches out whenever moving. Trying to pan/move view is virtually impossible, object jumps away quickly

At 10'000 units I see no major issues, though moving view is still finnicky

Could this have to do with graphics card performance? Higher end cards have a higher precision Z depth buffer and cause glitches less frequently?

kim holder

@DuarteFarrajotaRamos that sounds like it could be it. my previous card was a GTX 960 and i just upgraded to a 1060. my processor is an i7 4790, 4 physical cores at 4 GHz

i can navigate around that little man without much problem at 100,000 units, though it does jump a little.

Duarte Farrajota Ramos

Yeas, tested again in a new Blender instance and navigation is much smoother now. I can move view and rotate but it is visibly "jumpy", motion is stepped and not smooth as if snapping to a grid

kim holder

Duarte Farrajota Ramos

Still no visual glitches or Z Fighting. I'm on a GTX980ti, though I suspect Blender UI is probably running from a GTX 580

kim holder

no sign of geometric artifacts still

Duarte Farrajota Ramos

10:48 PM

yep, nice and clean, seems smoother than mine

kim holder

the file that's open is also pretty big

Duarte Farrajota Ramos

that impressive, mine is just a test scene with a suzanne

kim holder

i have 16 gb ram, maybe that matters

Duarte Farrajota Ramos

I have 16gb too, I'm more inclined to think it is somehow linked to Z Depth precison of the card, though I would have thought that VRAM availability aside they should all have the same buffer depth.

« first day (1877 days earlier) ← previous day next day → last day (2123 days later) »

all rooms

Transcript for

Jul13

Jul '1814

Jul15

A place to talk while we wait for our renders to finish. Site ...

cycles eevee modeling rendering scripting

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC