Recent Forum Posts
From categories:
page 1123...next »
Re: Hey guys,
animate1978animate1978 1257809295|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » Hey guys,

Since I have not had a whole lot of time to fully work on this I have been doing tests of the layout scene, of DOF really and yes I do have to say that it is quite slow - what I thought was DOF at first was only motion blur and did not realize I had not checked DOF in Mosaic…. so I am testing it now with different settings to see if DOF is really desired through the whole thing.

So far I can so no it is not desired. However I do want to do it when I can, because it does look nice.

Also I am trying to plan out scenes that do not require full rendering of every element in the scenes, such as stationary cameras where the background can be rendered once and everything else in the foreground is composited later on, this will make rendering a LOT faster I think.

But in general yes I am starting to get back into the swing of things as the remodeling of the old house is nearing completion, though it may take a week or so before I devote a lot of time. In the spare time I do have I will be doing something at least when I can. Dan and I are going to be working on setting up the layout template for all scenes very soon and we can begin blocking out camera locations and movements.

I also have to contact the Animux guys and see if I can get them to fully upgrade Aqsis to 1.6, it has been several weeks since I last talked to them but I do not think it will be a problem.

Re: Hey guys, by animate1978animate1978, 1257809295|%e %b %Y, %H:%M %Z|agohover
Re: Hey guys,
c42fc42f 1256992314|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » Hey guys,

I'm not sure, that may not work too well because depth maps can't be motion blurred. I'd prefer to fix the problem properly in aqsis, and I know there exist algorithms which make MB+DoF fast (just observe 3delight to see that it's possible ;-) ). Of course it may be quite difficult - I'll find out when I try it!

~Chris

Re: Hey guys, by c42fc42f, 1256992314|%e %b %Y, %H:%M %Z|agohover
Re: Hey guys,
DanielWrayDanielWray 1256990271|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » Hey guys,

How about using the motion blur and leaving the DOF until the end, If we can get Z maps out of Aqsis via the multi-channel feature we could just use the colour ramp and DOF node in blender's compositor, I know it'll not be making use of the DOF feature, but I think in some cases, if the render time is far to slow we will need to make trade offs.

By the way are we still using the render farm from that linux distro place? (forget the name)

Re: Hey guys, by DanielWrayDanielWray, 1256990271|%e %b %Y, %H:%M %Z|agohover
Re: Hey guys,
c42fc42f 1256966440|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » Hey guys,

Hmm, I'm not sure about why the extra eysplits warnings…

By the way, we've recently had some reports that using MB+DoF in combination is really slow and after doing some tests I can confirm that it sure is :-( So don't be surprised if activating both of those makes things render real slow in the current version of aqsis. I plan to have a look at fixing it when I get some time, which hopefully won't be too far in the future.

~Chris.

Re: Hey guys, by c42fc42f, 1256966440|%e %b %Y, %H:%M %Z|agohover
Re: Hey guys,
animate1978animate1978 1256959866|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » Hey guys,

Yes things are slow as molasses right now, mainly because I have been in the middle of remodeling a house so it can be sold, so that has taken up quite a bit of my time.

I do plan on getting back into this full swing in November (2 days? lol) and I will begin by at least making a rough animatic of this first.

From there I will be able to see what objects are in scene so I can remove the ones not seen by the camera. Even though the layout and animation will require some work, doing a rough animatic with the models will at least provide some idea of where to go.
I did notice speed increase! In fact I did a test render frame of an camera fly through, one using just motion blur and another with only DOF - while it took some time and a lot of complaining by Aqsis about eyesplits, the renders did in fact go a lot faster than expected. In fact the last time I did a test like this the render never started so that alone is a huge thumbs up from me!

Again sorry for the long delay and hopefully this will be the last big break so this can finally be completed.

Re: Hey guys, by animate1978animate1978, 1256959866|%e %b %Y, %H:%M %Z|agohover
Re: Hey guys,
pgregorypgregory 1256946495|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » Hey guys,

Hi Dan,

Count me in too, I'm still very keen to see this project come to something. As Chris said, 1.6.0 of Aqsis was released recently, with significant speed improvements, and of course the multi-channel EXR support that came from this project.

Looking forward to getting back underway.

Paul

Re: Hey guys, by pgregorypgregory, 1256946495|%e %b %Y, %H:%M %Z|agohover
Re: Hey guys,
c42fc42f 1256943311|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » Hey guys,

Hi Dan,

It's cool that you've got some time soon, I'm still on board for sure. I finally fixed the SDS speed a bit before aqsis-1.6 was released, so I hope you guys will notice the render speed with SDS has improved quite a bit :-)

Cheers,
~Chris

Re: Hey guys, by c42fc42f, 1256943311|%e %b %Y, %H:%M %Z|agohover
Hey guys,
DanielWrayDanielWray 1256927965|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » Hey guys,

Hey all,

Sorry for the absence, I've been extremely busy with university lately, Thankfully I get a week off very soon, which I'm planning on getting this project back into full swing. I haven't heard from ted in some time, I'll email him though and see what is happening.

Is everyone still on board? :)

Thanks,

Dan.

Hey guys, by DanielWrayDanielWray, 1256927965|%e %b %Y, %H:%M %Z|agohover

Cool, just so I know I've been clear you can still use delayed read archive on all visible objects and dupli objects. Just not on visible particle emitters without using "Particles in Bounds" and not on invisible parented dupli objects without manually adjusting the bounds (easier to just use read archive for these) ;)

Eric

Re: Recent Aqsis build render errors by Eric BackEric Back, 1253478518|%e %b %Y, %H:%M %Z|agohover

Yes I did some further rendering of the scene itself and all mentioned above does seem to be the case, which is no big deal just something I have to document for further reference. I dont think a bug fix is in order with Mosaic since after all it seems to be case by case basis, it is not something that cannot be fixed. Once the scene layout template is completed it will be far easier to work around these little issues. The idea I have is since a lot of the repeating objects are duplivert instancing it will be very easy for each scene to be custom tailored, removing vertices for objects that just will not be seen at all by the camera. Since the template is also nearly complete the other objects also fall within those rules. So placing the camera in a location will give a very good idea as to what will be included in the scene, of course if the camera is animated, again making note of what geometry is in view will determine what needs to be in the scene file.

So this is good progress, I was starting to think that something was very wrong with either Aqsis or the models themselves but in the end it turns out that it is not so bad after all and can be worked around easily. I will make sure to document these things in the wiki.

Re: Recent Aqsis build render errors by animate1978animate1978, 1253477131|%e %b %Y, %H:%M %Z|agohover

Hey all,

The problem is the bounds of each object does not include dupli objects. This means you should apply Delayed read archive to the objects being duplied but not the parent object who's geometry is dupli verting. What is happening is the parent object is setting a bounds around its geometry that is too small for the children. Disabling delayed archives for the parent object is no big deal since it's not actually exported only its transform ;)
I'm going to leave this as is however as I can see specialty cases where a bounds in a bounds may be necessary with dupli's. Another way around this issue is to artificially extend the parent bound to include the dupli's by using the "Manually Adjust Bounds" dialog in MOSAIC's geometry tab. To make it possible to "see" how far your adjusting the bounds you can enable the "Bounds" draw extra in Blender's Object->Object buttons toggles and MOSAIC will export a representation of the bounds ;)

Also note a similar problem can occur when using particles but in the geometry tab you can enable "Particles In Bounds" to tell MOSAIC to extend the parent emitters bounds to include the particles which by themselves have not bounds. However use this sparingly as some renders seem to render points no problem without bounds and this can seriously slow down export of particles systems and hairs (as MOSAIC has to cycle through every point position to extend the bounds by)!

Eric

Re: Recent Aqsis build render errors by Eric BackEric Back, 1253429075|%e %b %Y, %H:%M %Z|agohover
Re: Recent Aqsis build render errors
c42fc42f 1253417751|%e %b %Y, %H:%M %Z|agohover
in discussion R&D / Rendering » Recent Aqsis build render errors

Hi guys,

I just had a look at SCENE01.tar in the Dropbox directory, and it seems like the problem with geometry missing from buckets shown in the first few images of this thread is due to incorrect bounds for DelayedReadArchive, which is a MOSAIC issue I guess. (I notice that I get the same artefacts with 3delight as well.)

I cut down the scene a lot by deleting geometry until only the columns were present. In the file "Scenes/F1SCN-scene-DEFAULT.rib", the file "Archives/Objects/F1OBJ-Conc_inst_01-scene-DEFAULT-00.rib" is read using DelayedReadArchive. If I change that into ReadArchive, the columns seem to render fine, which to me indicates a bounding problem.

Perhaps it's something to do with bounding of instanced geometry? The file which is being DelayReadArchive'd contains lots of instances of the columns.

~Chris.

Re: Recent Aqsis build render errors by c42fc42f, 1253417751|%e %b %Y, %H:%M %Z|agohover
Re: R&D - Asset Management
animate1978animate1978 1253297592|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » R&D - Asset Management

Well….

Initial tests of Arachnid proved to be unpredictable at best, so Paul and I are looking into options. This may explain why Unison is no longer being developed, not sure but this week we tested it out quite a bit and it seems that after a period of time the transfers just hang, which of course is not good when trying to obtain the 850+ MB of data that has been collected.

We may end up using SVN after all.

Re: R&D - Asset Management by animate1978animate1978, 1253297592|%e %b %Y, %H:%M %Z|agohover

I am checking it out right now! Very nice!

So far I have done tests with just TIFF images and have yet to test the EXR, though it is next on my list.

I have made a tar of the scene 01 render example now and placed it in RIB_Bugs of Dropbox - Paul and I are still working out the kinks of Arachnid. One of the issues I did notice was Bounding Boxes, so I have kept that in mind and moved the camera to a new location. This is a large tar, I am sorry for the size but it is the only case that can provide the best instance of what I am talking about, I tried to reproduce this using simple geometry but have not been able to do so - thus the reason for the 4 planes and Suzanne test.

Then I did make this same test using an Aug 21 build of Aqsis and this too has reproduced the same exact rendering error, so now I am confused. I am going to be recreating the environment to see if maybe it is in fact the geometry location, the camera location and angle or whatever else seems to be causing this can be resolved. I am also not fully awake at this moment and apologize for my lack of wording.

Re: Recent Aqsis build render errors by animate1978animate1978, 1253119176|%e %b %Y, %H:%M %Z|agohover

Hey Ted :)

Just thought I'd let you know I've created two micro blog tutorials demonstrating the new features recently added to CVS (over two weeks longer then I said… sorry).
Here's a tutorial on the new volume shader (complete with video) http://sourceforge.net/apps/wordpress/ribmosaic/2009/09/15/volume-shader-rewrite-complete/
And here's a tutorial on the render layer system http://sourceforge.net/apps/wordpress/ribmosaic/2009/09/16/render-layer-support-complete/

Eric Back (WHiTeRaBBiT)

New CVS MOSAIC release with halos and layered EXR support by Eric BackEric Back, 1253084463|%e %b %Y, %H:%M %Z|agohover
Re: R&D - Asset Management
c42fc42f 1252898781|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » R&D - Asset Management

Any effort to avoid spaces in file names gets the enthusiastic thumbs up from me :-) Good luck with changing things over!

Re: R&D - Asset Management by c42fc42f, 1252898781|%e %b %Y, %H:%M %Z|agohover
Re: R&D - Asset Management
animate1978animate1978 1252883580|%e %b %Y, %H:%M %Z|agohover
in discussion General / General » R&D - Asset Management

Ok guys! I decided to NOT use ANimux network for the asset management, they have quite a bit going on with their own projects and whatnot, plus other reasons. Either way I asked my friend Ryan that owns the server we currently use for backup if he would mind just doing this part for us and he is ok with this. Though I do plan on using Animux for rendering, that part remains the same! I will make use of the SVN currently in place for this.

So the Arachnid project is finally coming into being! Tomorrow Paul is going to be workin on that server making this system work as we need it to and sometime this week it should be considered operational. So be prepared for the change. Between Dan and I, I think we will continue to use the Dropbox system for other purposes related to the project, not sure what at this time, maybe just as an extra location for WIP files or something - who knows?

That is when I noticed that the directory names are not Linux happy. This has been an oversight for months and I JUST noticed this. The problem was with spaces in the names, so while the majority of the names do not have this, some of the top level directories do. I am going to fix this up today, plus verifying that ALL assets are either unaffected or fixed. Then I will overwrite the entire directory tonight via FTP, hopefully this will not affect Pauls work that he needs to get done to get Arachnid running.

So from now on - no spaces in names at all. Use underscore (_) or minus (-), just no spaces.

Cross your fingers!

Re: R&D - Asset Management by animate1978animate1978, 1252883580|%e %b %Y, %H:%M %Z|agohover
Re: Recent Aqsis build render errors
c42fc42f 1252710851|%e %b %Y, %H:%M %Z|agohover
in discussion R&D / Rendering » Recent Aqsis build render errors

Hey guys, sorry for the slow response :-/

To me these issues look like bugs in aqsis, though I don't know what could have changed in the last few weeks to cause these. If you can make a simple RIB file for each of the problems I will have a look at them. Please try to use geometry which is as simple as possible, thanks!

~Chris

Re: Recent Aqsis build render errors by c42fc42f, 1252710851|%e %b %Y, %H:%M %Z|agohover
Re: Texture Map errors
c42fc42f 1252710166|%e %b %Y, %H:%M %Z|agohover
in discussion R&D / Rendering » Texture Map errors

Does it say specifically "Deflate compression support is not configured"? That error would come directly from libtiff. Unless you've changed the version of libtiff, I can't see how this error could suddenly appear…

Re: Texture Map errors by c42fc42f, 1252710166|%e %b %Y, %H:%M %Z|agohover
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License