Jump to content

chriskelley

Members
  • Content Count

    60
  • Joined

  • Last visited

Everything posted by chriskelley

  1. I've done a fair amount of research on distributed rendering systems. The biggest hurdles are things like: • Plugins (Using plugin X would require every render client has plugin X installed) • Plugin Licensing • Fonts • Bandwidth (Often the transfer time required for retrieving final renders can be untenable) • Privacy of files/renders (In the context of community rendering of professional projects) That's just a start. There are plenty of other debugging issues that arise when dealing with a heterogeneous network of machines that you don't have access to. But, it's certainly not impossible. Thinking about it is the first step!
  2. You can use the ease if you want, but your current setup is using it incorrectly. Try this: target_layer = "foo"; start_value = position; target_value = thisComp.layer(target_layer).transform.position; start_frame = 0; length_of_blend = 60; //in frames progress = (start_frame + timeToFrames(t = time + thisComp.displayStartTime, fps = 1.0 / thisComp.frameDuration, isDuration = false))/ (start_frame + length_of_blend); ease(progress, start_value, target_value); Documentation on expression interpolation methods can be found here: http://help.adobe.com/en_US/AfterEffects/9.0/WS3878526689cb91655866c1103906c6dea-7a06a.html
  3. Might want to try using Functions->Split instead - that way you can just break out all of your pieces and skip some of the delete steps. Other things that work on screwy normals of imported models are to: - Select all of your polys and run Functions->Optimize - adjust your phong settings - run Functions->Align Normals if some of them are getting flipped. Good luck Chris Kelley
  4. UPDATE: - Congrats to the quickest in the crowd, tickets are all gone. Hey everyone, I've been given the go-ahead to hook up some tickets to PAUSE: Ideas in Motion. Super easy to get em, just know a little bit about the history of your trade... http://themographblog.com/2008/11/05/free-...deas-in-motion/ Hat tip to FITC for hooking these tickets up, great to see them supporting our community in such a strong way. PAUSE - Dates, Times and Location: November 10 - 11, 2008, 9 AM - 6 PM BMCC Tribeca Performing Arts Center 199 Chambers Street #110SC New York, NY 10007 http://www.pausenyc.tv
  5. Thanks for the link up graymachine - just one thing to note that I don't mention in the video, is that Particular Emitter Z position has a max of like 30,000 (i think that's what i remember hitting...) So if you have some HUGE scene that you need to fly through, keep that in mind when structuring your scene in AE. Edit: I've only hit the max once since Particular was released, and I use it all the time, so suffice it to say it's rare c
  6. congrats on the licensing, keep up the good work!
  7. You could also add a wiggle() expression to your Focus Distance. Wiggle() still works if you already have an expression enabled, you just need it to happen after your first expression. For example: targ = thisComp.layer("layer1").position; cam = thisComp.activeCamera.position; distance = length(cam_z, targ); distance; wiggle(4,400); //the first number is the frequency, 'how fast' it wiggles, the second number the amplitude, 'how much' it wiggles Alternatively, you could also throw a wiggle() on your aperture like _gl said. If you don't feel like wiggle() gives you enough control, you can always set them by hand or use "the wiggler" to set them for you (Window->The Wiggler) hth c Edit: If you want it to be a more "steppy" look (notice you mentioned hold keyframes) you can add in a posterizeTime() to the expression, like so: targ = thisComp.layer("layer1").position; cam = thisComp.activeCamera.position; distance = length(cam_z, targ); distance; posterizeTime(12); wiggle(5,600); Also, remember your Aperture will come into play no matter which technique you use, you will want to adjust it to get the proper amount of blur with your wiggle.
  8. graymachine I know you like to get loose with some code -- I think there may be a COFFEE solution in here that includes (blue sky thinking here stay with me) setting up something that would raycast onto your model from a set of spherical points/angles, and storing any object that gets casted upon into an array, and anything that doesn't gets deleted. Probably too convoluted of a process for this one model, but if it's something your planning on doing often it may be worth thinking about further.
  9. yes pre-comping can be your friend on this one. I would cut your mask of the layer, then pre-comp with "leave all attributes" checked, and then go into the pre-comp and paste your mask in there. That might do the trick for you, not withstanding any continuously rasterizing issues that might rear their head.
  10. chriskelley

    "Floaters"

    There's an interesting wiki article about them: http://en.wikipedia.org/wiki/Floater Don't worry, you're not alone.
  11. Meh, pretty sweeping generalization. Also, if a studio has 15 machines, and the newest top-of-the-line mac is ~$7k, I think it's bad business for small-medium sized shops to spend $105k+/year on new machines year after year. I agree that shops should have *high performance* machines, but that doesn't necessarily mean brand new. I've worked with some quality shops in LA that still have some PPCs laying around, and either way they should always hold on to old hardware as long as it works... throw that baby on the farm!
  12. Depends how much they are paying you to update it each time and how often they need it updated. Say they pay you $100 to update it, and they need it updated 1 time per week. So assuming the life of the package is 1 year, they would be spending $5,200/year on getting the element updated. Their goal by buying the source from you is to not have to spend that $5200... but the problem is, now that's $5200 that you aren't going to make. So the bottom line is, you've got to find a happy medium that takes those two things in to account, along with the "keep the client happy". Also, if there was never an initial mention of source files in the contract, that should be taken into consideration as well. Based of the $100/update estimate, I would say $1000 would be more than fair to them - that's only 20% of what it would cost them over the course of the year. I would personally vote for 25% of the lifetime of the updates. ($1300 in example case). (cost_of_update * (frequency_of_update * life_of_project)) * .25 Stand up for yourself and your pricing - business is business. Do your charity work for the kids, not your clients.
  13. Still pretty new but I've got a few up over at http://themographblog.com - they're usually pretty specific issues that I cover but you may find some useful. c
  14. I updated Lloyd's script to remove the email routine. This one should work for you, worked fine when tested on OS X 10.4.1.1 http://chriskelley.tv/misc/Render_Save_Shutdown.zip
  15. Not entirely accurate -- every machine does *not* have to have the files on it - they just need to have access. If all of your machines are on the same network, and the files referenced in your AE project are all pulled straight from the server, then you only need to copy your AE Project file into the AE Watch folder for the render farm to work. The plugin registration is certainly true, but that is usually the case. Some companies (trapcode) provide special setups/registration options to ease this process. As far as random() / wiggle() goes, that's a valid point... I imagine you will have much better luck with any random calculations over a homogeneous network then you would a mixed network. AFAIK different chipsets implement randoms() differently, so you could run into some issues there. But to answer the original post, I do not know of any well implemented online AE renderfarms... always seemed to me that by the time you have files collected/uploaded/rendered/downloaded it would have been faster to just render the damn thing on the G3 in the corner.
  16. ooh this sounds pretty cool... I've never heard of it but I like the sound of it.
  17. If you want to tweak the spatial tangents of position keyframes, the way I do it is by using the pen tool (or option+drag) on the keyframes in your actual viewport, not in the timeline.
  18. Here are some pretty decent discussions on the topic, along with other options: http://lifehacker.com/software/lifehacker-...ices-298573.php http://lifehacker.com/software/backup/ask-...vice-180556.php http://www.techcrunch.com/2006/01/31/the-o...e-storage-gang/ Haven't used any of them myself, but it's definitely a competitive market right now so I bet there are some great deals.
  19. I would think sampleImage() would be the winner here, this technique from Dan Ebberts is the theory... http://www.motionscript.com/design-guide/obscuration.html Except instead of alpha you need luminance, which can be derived from the RGB values given back from sampleImage(), though I don't know how off the top of my head, graymachine may know. Also you would need to change your sample area to whatever area you want averaged on the strobe footage.
  20. Very nice Adam, very nice indeed. Great utility.
  21. ooh, nice solution Harry, works great. Only thing is that every once in awhile it pulls frames twice in a row, which would be difficult to prevent, but good job.
  22. The problem that arises is that you can't access the "last random frame generated". Self-referencing inside expressions is a big headache. It's easy enough to do your time offsets and timeRemapping to random frames, the problem is that after you generate a random number to use for your timeRemap value, you want to hold value generated by random(x) for y frames... but as soon as you hit frame random(x)+1, you lose reference to what the value of random(x) was. The debacle is summed up pretty good here by the godfather of AE expressions: http://forums.creativecow.net/readpost/227/10157 So bottom line, unless I'm missing something, you will want to adjust your workflow to not use expressions for this technique. Maybe try generating your keyframes with wiggler like dachavez said, and then you can adjust the "hold length" by selecting all your keyframes and giving the last keyframe the ol' "alt+drag" stretching technique to adjust the gap between them.
  23. I tested them briefly last night, didn't have any problems. C4D, Client, and Server all have the same icon visual, but don't use the same icon file... so there are no conflicts as far as file usage goes. Any other stability questions would stem from customizing icons in general, which though there may be some .001% chance of problems, I would imagine there are inherently very few issues with it.
×
×
  • Create New...