Forum Replies Created
-
Posts
-
I guess you are talking about the above image produced with inferno.You can get the batch setup from:
http://oyoon.home.comcast.net/shadowprojection/shadowprojection.htmThanks to Danny Yoon, for his set of 3d inferno hacks. Danny showed us this at the year 2002 with inferno 4.6, and we still want to see see this in an upcoming version of inferno.
The new fusion features seem great, for a minor upgrade. Congratulations to Eyeon.
merv wrote:cool…what is N+L by the way?…cheers…merv…N: Nodes (which are your axis, images, geometries in action schematic)
L: Layers (which have blur, crop, CC, keyer setups)If you don’t want layer setups, you can just select “Add N”, and it will load just the nodes in action schematic.
or you could just add the second action setup to the first. just select “add N+L” while loading the action setup, and the loaded setup will be added to the current. Copy/paste the channels you want, and delete the ones you don’t need. This way, you won’t need to use batch.
TomD wrote:At the risk of really sounding like I don’t know what I’m doing, where do you find a median filter in Flame?On desktop go to “processing->filter”, and also in batch you have a “filter” node, or you can have custom filters, including “median” in desktop paint module. Use the method whichever fits best to your requirements, they all do the same processing.
My additions would be these:
I would also look to antialiasing samples, if you are scaling down your titles somehow in action. You will need more samples, as you scale down your images.
But the best way would be, not scaling any of your images, and rendering with 1 sample only. And if possible, move titles in discrete pixel counts/frame, then even turn of filtering in images, to keep images as sharp as possible.
If you move your titles too fast, there is no way to deal with 35mm projection strobe. So do not move it fast.
If possible, I would shoot with HDCAM to get rid of the fields. That will save hours in post (just because you have to deal twice as many frames, when working with interlaced footage), and daily rate of HD camcorder should’t much more expensive than a digibeta. You can shot true progressive images with HDCAM. But you will need an HDCAM VTR and HD video input.
I would definetely go for HDCAM, also for resolution reasons, you will get very good keys, when you downconvert HD to SD resolution. You will save lots of hours in post + you will get much higher image quality as the result.
osmose wrote:Working with proxies isn’t dangerous ? I mean when rotoscoping etc, does the “interpolation” of what you did on the proxies goes well with the full 4k. I’m afraid to have a lot of bad surprise when rendering at full 4kjulien
Quantel tells you that proxies are always bad. That’s not true, they are sometimes good, sometimes bad. For masking, tracking, keying, etc… we work in fullres. But since tracking is done in one layer, even an onyx2 is fine for 2K tracking, and also the same with masks. But when you want a quarter res preview for your 40 layer comp, to control timing, etc… a proxy render might suffice. At those positions, proxy workflow is great. But for controlling your key, tracks, you always have to check them in fullres, if you want to be precise.
About Quantel systems, we have been demoed just today for their Pablo system. It is fast, I mean very fast. Works great in 4K! A primary color correction, with 300-400 pixel blur renders 5-6fps!!! But if you ask my idea, it is very fast for color correction and conforming, and for very simple compositing tasks, but it is not good at complex compositing jobs. It doesn’t have a true process tree, and it lacks the tools for highend compositing.
I have never seen an inferno on linux yet. It should be a hell faster than the onyx systems. But it might not be as fast as Quantel dedicated hardware…
4K needs lots of bandwidth, approx 1.2Gbyte/sec!! Even for doing a ram preview, you will need 1.2Gig of system ram for each sec. This will physically mean 6sec. of max length of ram preview for 8GB systems.
I can only tell my limited experience with flame/inferno systems, 4K doesn’t work well in 32bit versions up to inferno6.2/flame9.2. So if you will ever use inferno for your comps, I would try v6.5 (64bit) and 12gig of ram on linux system. I hope you can comp 4K w/o any memory problems in batch on that version.
2K proxies?!!! We work here with 720 pixel proxies on inferno for 2K! 2K is damn slow for interaction, if you have more than 2-3 layers!
Regarding combustion, does it have 64bit version? If not, 2GB will be your limit with that application, and I don’t think it might be possible to do complex 4K comps, with that amount of memory. Don’t know about DF has 64bit code…
The 64bit apps I know: shake/inferno6.5…
About 4K projector, Sony has one. We bought sn0001 and went into some problems, Sony tries to solve it now. And it is not true 4K! It has 4 x HD1080 inputs. It should be priced around $80k, if I don’t remember wrong. But even if you buy that projector, to drive it at so called 4K, you will need filmlight system for playback. That was the only system supporting such a config.
Shadows are really a pain in action! I sometimes create another layer for the shadow, because I don’t want to go for shadow mix (just because of Z positioning issues).
I also guess that it is the AutoCC, try to play with the range value, until you find a good spill suppression amount.
On still camera shots, you can try compound menu. You can compound 50 frames into 1 frame for example, if your camera is still, and you can try masking the moving objects.
We work at 48Hz for 24psf projects on our onyx350. Which we can work with for hours, and even days. Maybe we got used to 50Hz so much, because we use it everyday for our PAL work. Maybe you have another problems… Do you have a flourecent light in your room?
Have you tried to use z-sort button in layers menu?
Some funny requests/questions from clients:
1. Client: Could you please add a light over this background?
Me: But the background is fully white?!2. Client: Up to which point can inferno scale the video, without distorting it? We can scale up to 30% in avid..
Me: Let’s scale up to the point, that you don’t feel any distortions.3. Client: Will this text move so slow? I thought you would make a faster move.
Me: No, I was just rendering the setup, I will play it back, after the process finishes…4. Client: What effects do we have in inferno?
Me: Inferno doesn’t have any effects, we just do fx by layering the materials you give me.5. Client: Ohh, you don’t see an image on the tablet. But how do you paint it, without seeing anything on the tablet?
Me: We just get used to working with pen, while looking at the monitor 🙂6. Client: We didn’t bother with the talent, during telecine. We just have concentrated on greenscreen.
Me: So you have done secondaries during grading for the greenscreen shots?7. Client: We have printed our HDCam shoots to film, and transferred it back to video, to get filmlook. Did you like it?
Me: (Could I say something logical?)These are the things I remember for now. Ofcourse there are many other things that I don’t remember. But why don’t you post your ones to this thread?