scott8933
-
Posts
33 -
Joined
-
Last visited
Content Type
Profiles
Forums
Calendar
Posts posted by scott8933
-
-
I recently found that After Effects has a great motion blur, but it's NOT the Frame Blend effect you activate in their Timeline. If you open the TimeWarp effect, and set the speed to 100%, it won't function to speed up or slow down a clip. But, buried within TimeWarp is a channel called Motion Blur. Twiddle with its values and you can add great motion blur to any clip. Why they bury this function so deeply and keep their lousy Frame Blend is a mystery to me.
Joe T
AE's stock mblur has gotten a little better over the years but is still pretty crappy for high blur settings. The TimeWarp filter does the same thing, its just a step-frame motion blur. You can get much better results in either by upping the sample steps, controlled with "Shutter Samples" in Timewarp and "Samples Per Frame" in the Advanced tab in Composition Settings.
They look similar though slightly different (I actually see more banding with timewarp), doing it through Comp Settings renders far faster though. I have a feeling Timewarp is pretty old and unloved.
FWIW there's also an mblur setting in the Transform effect, which will override your comp's blur value, though its also a step-frame blur and has no inter-frame setting.
-
Some plugins behave very strange with mblur, for example one of the Northern Lights plugs will have extreme blur coming from the center of the plugin group's coordinates - as if its taking every object and shooting it out into position for every frame. Don't remember which one, maybe Placer? (Also this might have been fixed at some point).
The extreme blur settings have always been a way to get effects like the ultra-streaky/speedlines look, i.e. in The Mask when the main character spun-transformed into his Mask alter-ego.
This works in After Effects as well, btw.
-
Given more time I'd like to have figured out the solution for the Lite build, since not everyone has the option to switch to the full version... and actually I like the easier-to-use Lite option panel. But as long as it works I'm happy.
It seems like it would be easy to analyze since it was "random" (in time) but completely repeatable. Also, I forgot to mention but just remembered, the dropout was effecting all visible geo in the scene, not just one or two models, and I'll have to double check but I think it did it when I replaced a model with an Ubershape as a test.
-Scott
Ola Scott,
Great you found the problem, a simple new NPR Full build :)
Thanks
Tom
-
Problem fixed by switching to the full version of NPR; I would have preferred to have figured out what was causing it, but no time.
Things I tried that had no effect:
-Change camera position and angle (slightly)
-Change light setup (more lights, less lights, change light position, change intensity)
-Change object's luminance, diffuse, specular
-Change settings in NPRLight
-Change frame rate from 24 to 48
-Change from Raytrace to Phong
Every time, same result. Every 20 or so frames (at random, no pattern I could see) either all or some of the shaded area would not shade. Next frame it would come right back. Sometimes only half the frame would drop out, and it appeared to correspond to the bucket size but it was rendering too fast to be sure. Camera was moving very slowly, no sharp movements in the scene.
When I changed to 48 fps, it dropped out at the same frames and occasionally the dropouts would extend for half a frame before or after.
No more time spent troubleshooting.
-
I could understand (and fix!) if the problem were consistent, but the camera is moving very slowly across the scene, and the flashes appear out of nowhere for just a frame. Not always on the whole scene either, sometimes just the lower half. Might be related to the bucket size? I'm on a borrowed box with not much ram, I haven't actually changed Camera's stock memory settings.
I tried a few things, changing lights, changing the camera angle slightly, etc. It always pops at the same frames.
Solution 2, I tried doubling the frame rate (if nothing else I can frame blend the shadows with this pass and minimize the flashes) and it still did it during the same frames. Again sometimes the lower half only.
Hello Scott:
Dont have a easy solution, NPR get the normals info and paint the perpendicular poligons to camera with shadow zone. No same position all time.
If can add more poligons with dicer can solve a bit...
Thanks
-
Hello everyone,
I'm using the NPR shader to do an outline render, it has a little bit of solid shading on the edges of the objects right now, which is fine except that it flickers throughout the animation. Flickers, as in parts of it (in strips, inconsistently) will just drop out for a frame then reappear as normal in the next frame.
I can probably blend it out in AE, but I'd rather have it render correctly in the first place. Anyone know the cause and solution?
The attached files show the correct frame in all black and the next frame with a portion dropped out, highlighted in red.
Thanks.
-
I'd actually bet that the old 1.0 interface looked a lot like the 1.5 interface (which was the first version I used), which honestly looked pretty much like the same one we have today. It hasn't really changed a whole lot, other than more "polish" and many more features.
A couple years back I found a version of Photoshop 1.0 and was able to run it - guess this was pre OSX - and was surprised to see that it looked very much like the same Photoshop we use today. Just with fewer buttons.
(not my screengrab, just a GIS)
Though when I moved into a new office, we did have some pretty old apps including this gem:
Chip,
What does it say about me, when an old floppy disc really gets my blood pumping on a Monday morning? Don't answer that. Really fantastic post. I've never seen one of those puppies. So cool to see. I wish I could see and dive into the interface of the actual software somehow. That should at least be framed in a quality box and hung on your wall. I'm a little depressed that most anything I've done since doesn't live up to your space station project rendered on a 10 MB hard drive back in the early 90's. I started 3d modeling in `92 on an SGI. Props to you, Chip. That is really cool stuff.
Keep `em comin EIAS pioneers and fanboys! ; )
-
Back when I worked at ABC On-Air Promo, they had a copy of the docs for a pre-1.0 version of EIAS. I don't recall if it was referred to as "Spotlight" still (don't think so), but they were just xeroxes that were spiral bound. Never saw the install disks for them, by then we were using a newer version so the docs were just filed away. In the Paintbox room, of course.
That room was always literally and figuratively dark. The Paintbox hadn't been used in a few years and just sat idle, though turned on. Its control was downstairs in the machine room (nothing was ever turned off down there). At the time we were still running a Harry, so the Paintbox was redundant so to speak.
3D was mostly done in a different office where they were running Prisims and SoftImage. None of the designers besides me did any desktop based 3D, since the full-time 3D guys were really top notch and their SGI's (Indigo and something else I forgot) were comparatively faster.
The ADB dongle had a very low serial number, under 100 as I recall. I thought about keeping the docs just for posterities' sake, but they were really more of a curiosity than anything else. The EIAS manuals were notoriously uninformative back then and these were no exception - just much shorter.
This seemed pretty typical though for CGI at the time. Once we upgraded the Harry to a Henry, we got new manuals for it which were only about an inch thick. This, for a machine that cost upwards of $1mil at the time ($500k for the Quantel, and another $500 for the D1 you had to hook it up to, though it would work with most any kind of tape deck.)
I sure did hate those D1's - they rarely worked right, having some 20 heads that had to stay aligned - so half the time you'd have this really expensive piece of media that could only be read back by the machine that wrote to it in the first place. But ABC kept really good libraries and records of their stuff and there was an entire room with several hundred D1's, along with 1" tapes and what not.
All useless and unreadable less than 10 years later - millions of dollars of hardware and artwork, all basically scrap.
Wait, what was the topic of this thread again...? Something about iffy career choices, I think.
-
Minning also makes a companion plugin for doing toon shading:
http://www.minning.de/software/celulight
I haven't used it, but have used Normality with much success in the past.
-
Having just finished up something vaguely similar to this, I'd say you may want to try rendering in layers and creating the look in post.
The project I was working on needed to have that Sin City hi-con look, but technically this has similar problems. Doing it in-camera is going to lead to crawling/boiling/etc.
Our solution, not the best, but it worked - everything was rendered as passes with a fairly simple color and shading palette, then the Maya guys had a script worked out that would render matte passes for all the parts we needed to effect differently (if necessary). This allowed for a ton of control in AfterEffects, both scene-wide, and then on a part-by-part level by using the mattes. There are plenty of good techniques to get various looks in post. We were mainly just using levels and curves to hicon things selectively.
One of the guys on the project had actually worked as a lighter/comper on Sin City, and he described some of their techniques - lots of projection mapping onto geometry. This worked quite well, and could work for your project too. Requires more prep, but I think ends up with a simpler pipeline.
Hi Guys,
I'm working on a piece that requires 2D (comic frame by frame other artist) and 2d cell (3d - me) and regular 3D backgrounds (me). I was wondering if any of you have animated something similar to this:
Animatrix -
http://www.youtube.com/watch?v=exEO94Nflng&feature=related
or
(disregard the sound track)
http://www.youtube.com/watch?v=83O79Iq8jRc&feature=more_related
I have the DVDs for this series and my understanding was that XSI was one of the main3d apps with 2d cell rendering used.
Or the Gorillaz
http://www.youtube.com/watch?v=pw8PpYBiDsc
(guitar closeups and drummer)
If you have animated something similar, my question is for modeling and rendering. On the modeling front is it safer to model a character like SouthPark / paper cutouts? Or is it fine to model the full 3d element with volume whereas the 2d cell shader and outline assists in the look. How much of the look is in post? ( I know I saw a sketch robot in the gallery that has blue print turn into 3d with 2d cell, curious about the techniques used in that one)
In some cases it seems like an after effects 3d layer composite.
In the animatrix case my question is the cell "look" that the 3d objects cast in the beginning of the pyramid down to the street level and in the second clip the camera flythrough of in the begining showing the army and the robots (the robots cell / phong looking shader with outline on).
What additional shaders if any are recommended for EI, or what plugins do you all recommend that may ease producing this kind of effect?
I'm sure you all have busy schedules but any tips are very much appreciated.
Best regards!
-
I haven't ever had time to really dive into the photon features of EI, but as a compositor I've been getting AO passes from Maya for quite some time - they usually do a pretty nice job of faking GI, sometimes even contact shadows - though they tend to be pretty noisy, sometimes bad enough that a secondary filter is needed to kill the inter-frame buzz they make.
How does EI's Photon method differ? Better quality? Slower or faster? Used for different applications? The "maya" method is to dump several passes into the comper's lap and adjust from there. It has its advantages and disadvantages of course.
I'm just wondering if the EI Photon process would be able to compete (hopefully exceed!) a Maya AO pass for fast faking of GI. Or do I have it wrong in its purpose?
Ola my friend,
“It´s the same "Sala Test" that i uploaded for you some time ago. My theory is if i can bake the lights (whatever time it takes) the animation after that will be faster (a pan in to 5 boys and girls chitchating near the blackboard during 8 seconds) if the Gi is calculated for every frame, (and i can´t exclude them) i think i will eliminate the Gi...“
Look my attached, the project bake light with Photons in the geometry, but dont bake GI, GI is disabled.
After the first frame, which is slow, each frame of the animation is rendering in 2 or 3 secs per frame (Photons Calculate pop-up as Permanent).
The only negative thing is, this approach dont have the richness the GI render quality.
“Can i bake the lights for the empty room and, then, ad the children iluminated separatedly (not baked)? i m not sure if i make me clear...:huh:â€
You could do your full scene, then render in 2 passes with a mask your children, render the class, then the children.
I can help you to setup later, if you want.. :)
“FelixCat
Sorry for taking your time, i think Ei 9 is taking all your time right now (and your real job too)â€
Im having a lot of pleasure, because we develop the tools to make our lives easier here in our community including me which I use every day.
Happy New Year
Thanks
Tom
-
A good game card will work perfectly with EI, you can get a nice nVidia one for $150 or so that will have dual DVI out and about 1gig ram onboard. ATI cards have traditionally not been as compatible, though I've got an HP that came with one that ran just fine... before I replaced it with a GeForce.
The higher-end Quadros and Fire cards are a major hassle to run, I found a previous system became less stable when I switched from a top-of-the-line game card to a high-end Quadro. With the workstation cards you have to be much more careful about which drivers you're running. Everything gets much more finicky.
I noticed no difference at all in After Effects, btw. I was able to turn on more effects in the OGL prefs, but previews and scrubs were just as atrocious as ever when using OGL Preview modes.
With a good game card, you can still turn on AA settings on an app-by-app basis - so you can still get better looking views to some extent.
Hi all
I want to change my graphic card. I am on PC machine.
do you think it will be better with a C.A.D. card like that : ATI FireGL V7700 compare with a same price game's card ?
thx
- -
psl
-
Well, at least in the post side we have more options than ever; Illusion has been made into an AE plugin, and Particular has been ported to Nuke.
Any news on Northern Lights? Blair, you're kind of the Last Man Standing.
-
http://www.thefoundry.co.uk/articles/2010/11/03/167/mari-11-on-windows-is-released/
Not quite the Photoshop killer I'd been hoping for - its very much tied to a 3D texturemapping workflow. I didn't play with it too much yet, but it seems to insist on importing an .obj before it will even do anything. So; great new texture map tool out there! But unfortunately, those of us who are still looking for an alternative to PSD for general painting may still need to keep looking.
Anyway, looks like a great tmap application and the price is great. Mac port coming soon...
-
That's disappointing news; Dante was a pretty robust system before, and PPP was good but really not nearly as powerful as Dante.
Has Blair weighed in on the stability problems?
UB Dante is really buggy, I used it a lot in EIAS 6 on a PPC without issue. Since version 7 and the UB version was released it has been unusable. Even when you get it to work it will corrupt projects to the point where they won't open until you remove the plugin from the Sockets folder and delete the object. It will crash EIAS consistently and constantly get memory errors.
The worst part is that it won't work with Renderama at all.
Try Power Particles Pro. It's has some limitations compared to Dante, but at least it works.
brian
-
Most of the NL plugins are pretty specific with how the attached groups need to be named - you had "autoname" turned off in uber, right?
I've used Dante with ubershape before, and it should work fine. It's safer to save out the ubershape as a baked model, but not totally mandatory.
Maybe post the project? Sounds like a simple thing to troubleshoot.
-
Your max reflection would be a 100% white area no matter what, so you'd have to fake it with glows/bleeds/halos/etc. AKA, the screen only goes that bright - so a reflection can never be brighter except by using an optical illusion trick.
I think what you're asking is this - you want an object that's not very reflective, but there's a really really bright thing in there to create a hotspot on said object. Without actually testing this, a couple solutions come to mind:
-You could make several reflection-only objects, all 100% luminous, and hope they somehow add to each other (probably won't work)
-You could texture-map a fake reflection onto the object instead.
-You could put a second object in the same place as the "real" one, and set it to be 100% transparent except for the reflection you want to use.
-You could do the same thing, except render the extra reflection object as a second pass and composite it in AE. That would offer the most control since then you could add glows and whatever on top of it without effecting the rest of the scene.
But luminance can't go above 1 either, so would that allow for bright reflections?
I want to be able to, for example, set reflectivity to 0.1, and have a bright white reflection from the light... As you can get using HDRI reflections.
(Can't find my dongle, and I've been too busy to do a through search... I really hope we can have software keys sometime...)
-
In one case, the odd triangulation thing worked out pretty well - I used it for the end graphic in the "Borat" trailer.
-
To us users though, that's sort of confusing - you guys picked up what looks like every other Konk. plugin, so my first thought was "oh, maybe they just forgot to put that on the list."
Is there an official word on Konkeptoine? Is David A. still making plugins or supporting the few remaining ones that haven't moved to the new owners?
Ola Scott,
David Argemi was the Fiber Forge developer, so, only him can bring it back.
Thanks
Tom
-
Wouldn't FiberForge have been included in the Konkeptoine set? It had some really useful features, like wind and motion reactivity. Would be a shame to lose it.
Ola David,
I always used Placer to my tricks with Fur, grime.., I never tested Fiber Forge.
We dont have a Hair solution now built-in in EIAS, maybe in the future? :)
I did with Placer.. look in the end of thread..
http://www.eias3d.com/forums/showthread.php?tid=322&highlight=Mouse
Thanks
Tom
-
Its been a while, but can't you augment the glow using the "fog" on top of it? There's an additive setting as I recall.
And if that doesn't work, you could just place two lights in the same place. Or just do it the really old way, and place a sphere in there with a falloff shader on it to simulate the whole thing.
Ah, sorry.
I mean for lights with glow... The round glow you get with spot lights.
Do you mean with object glows? Are they reflectable?
At the moment if you have a light with full glow and you want to reflect it (to get a nicer looking specular) unless the object has pretty high reflect settings (which will obviously reflect everything other than the glow as well, more than I'd like), the glow's reflection looks pretty dull.
I'll make a mock-up when I can find my dongle... Stupid thing.
-
Hah, all these years I never noticed that "start time" setting before.
Thank you guys. If the simulation looks good at 20 sec, simply put -20 in "start time, s" in the plugin setup window. Also remember that subdivide none gives you a nicer mesh than subdivide 3 steps. I don't think the Flag plugin was ever intended for closeup shots and it is far more slower and complicated to control with a dense mesh. I never tried with Encage nor Dicer. SimCloth can also make nice flags and looks nice at close up.
-
Richard,
Nice reference for Flag! Are you able to get it to render without the triangle problem when close up?
Christian,
One thing you always have to remember with Flag is that it requires a lot of pre-roll before settling into a steady state, and the pre-roll is cached separately between Camera and the Preview window.
I.E., you run the simulation in the preview window for say, 30 seconds. By then it should look like a nice Flag-Waving-In-Wind. Go to hit Render however, and Camera won't render the same frame. So when you find a look you're satisfied with in Preview, remember that you'll have to let Camera render just as many frames for it to get there - again usually about 20-30 seconds worth.
You'll have to start rendering from frame 1 to get there, so every time you do a Flag animation, take into account that you'll need a lot of pad at the head of your animation that will be getting thrown away.
If you plan on integrating a Flag into a bigger animation, you'll have to account for that - remembering that the pre-roll must be rendered for it to work. You can't start rendering 30 seconds in, since the simulation starts from the first rendered frame, and not the first keyframe in Project.
Also, unless its been fixed, Flag tends to get this weird "triangle" look when rendered. As if its not meshing enough or if the smoothing angle isn't set high enough. It looks fine at a distance - see Richard's example - but in close-up it won't look good enough for critical work. It helps to use a combination of Dicer and/or Encage on it, it also helps to use good (like actual photos) texture maps in the Diffuse, Bump, and Transmission channels. They won't fix the smoothing completely but will cover it up enough to sell it as a real flag.
Bonjour Christian!
Here is a little movie I made showing different flags values.
http://www.rdn.qc.ca/eias/Flag_References.mov
Attached is the Long Flag I made awhile back.
I hope it will steer you in the right direction.
-
Very nice clip! Some sound will really make it come alive. Had you considered a little post-work, like some tire smoke, light flares off the headlights, stuff like that? Would bump it up another notch.
Here's a slightly related link, its a custom Cobra made for Larry Ellison. Several million dollars to build, quite a feat of engineering when cost is no object:
Any hope to open a 1.3 gig .FBX?
in EIAS 8.0.0
Posted
Hello everyone,
I've got a 1.3 gig .fbx file, but so far nothing's been able to open it. I've been told it originated in SoftImage (really? is it still 1997? anyway...), and so far I've tried Max, Maya 2011/OSX, Cinema/OSX, and of course EIAS/OSX.
So far everything has either crashed or processed it for a long while then opened a blank scene.
AFAIK there's no animation or rigging, but probably some very hi res textures and the model is most likely a crazy dense mesh. Looks like a single object though.
Any ideas would be most appreciated!
Thanks.