Forum Replies Created
-
Posts
-
Folks at Nekochan reckon it’s a DM2 – http://forums.nekochan.net/viewtopic.php?t=7781.
Dive into the Quicktime bit in Control Panel, and in the Advanced tab, have a play with the settings in the Video section. Does it work if the draw method is set to Safe there?
That’s all very well in a studio but running an FCP rig on location seems a bit silly 🙂
To be honest… the XL1 is a great camera. I’m not sure you’d get better results without spending a whole wodge of money. Are you happy with what you’ve been shooting so far? In what ways would you like it to be better? Have you tried greenscreen keys already?
The XL-H1 looks lovely. But in order to record that uncompressed SDI out in the field you’d need to spend another whole wodge of money, so it’s basically similar to the Sony FX1 HDV camera or whatever it’s called, only with better lenses and ergonomics. However the SDI out should ease getting the footage onto your Onyx a WHOLE lot – image sequences over the network is way slow 🙂
4:2:0 video contains the same amount of colour information as 4:1:1, just arranged differently – the chroma planes are half the size both horizontally and vertically, instead of being a quarter the size horizontally and full size vertically. In fact regular PAL DV is 4:2:0.
HDV uses a bizarre sampling format, it’s like 17:6:6 or something ridiculous. It’s worse than 4:2:0 anyway. The increased resolution may offset this, though.
[edit: it’s actually HDCAM that’s 17:6:6. HDV is 4:2:0. My bad.]
mdillender wrote:Also, I was talking to someone today who works in the broadcast industry who said your better off bringing video into the Camera via an S-Video connection from camera to PC due to the compression via Firewire being to high at 5:1.Is this true and can anyone collaborate. Again, I’m trying to work in combustion and get some better keys.
That’s complete bollocks. It may key better if you do that, but that’s because the analogue connection smoothes out the chroma where the digital connection doesn’t. You can achieve the same effect with the plugin mentioned above, or this tip: http://www.fxguide.com/fxtips-223.html, and get better pictures to boot.
3 CCD cameras will give you a better picture, but basically the more you pay the better you’ll do. Keying DV is never going to be perfect. What kind of problems are you having? Steppy artifacts along edges, or what?
That Sony panel looks to have analogue HD15 inputs too, and it supports sync on green, so all you should need is a 13W3 (the weird SGI monitor plug) to HD15 adapter. Generally ones designed for Sun rather than SGI also work, there are piles on eBay.
Since the connection is analogue the panel will have to do its magic auto-adjust thing – it will probably do a better job if it’s got a sensible image to work with. I find the convergence test image from the IRIX confidence tests works well. Run /usr/sbin/confidence from a shell and do the monitor one.
That looks pretty nice. Did you just polar-coord transform the background as use that as a reflection texture? Are there any actual lights as well?
I tried the Paint-in-Batch thing – it doesn’t work quite as well as I was expecting, you seem to have to switch viewports to get the context viewport to update, but it’s still much more interactive than flipping out to desktop Paint. You can pipe the texture image through whatever other processes as well, of course.
If you try and import something really huge, wait a while and it should tell you in the console window what the maximum frame size is. For me it says 33553920, which I assume is in bytes, which gives you about 3344×3344 at 8 bits per component RGB. Try Soft Importing if you can, I just managed to do an 80 megapixel (!) .jpg soft import but I couldn’t output it from the colour corrector because the frame was too large again, you might be able to pan around it with Resize.
Those environment-map-as-shader tricks should get a whole lot more fun with Paint in Batch. I wonder if the texturing updates as you paint or only after each stroke…
Warps are soooo passé now 🙂
No, seriously, you’re right, it’s long and boring. The secret I suppose is to use as few source keyframes as possible. How is the new Distort? Has it surpassed Elastic Reality yet? 😉
If you mean satin as costumes for talent, then it can be a problem since it’s reflective and will pick up lots of the green which diffuses off the backdrop. Depends a lot on lighting, but if you end up with green highlights/sheens on shoulders and arms obviously the keyer might pick them up. If you’ve got a really well lit greenscreen you’ll probably be able to differentiate though, the green from the screen should be a lot more saturated.
Checks and stripes should be okay as long as the video never goes to a composite format.
HD D5 is intra-frame compressed at 4:1 or 5:1 and is 4:2:2 so it’s not going to be ideal but you should be able to key fine if you blur the chroma first, like this: http://www.fxguide.com/fxtips-223.html.
Good luck!
Short answer: no.
Sometimes you can do some extraction if the voice is in the middle and the music has lots of stereo content. But it’s a pain in the proverbial.
Does anyone know why the default is the “wrong” way around anyway? Does it relate to how actual celluloid-cutting editing is done or something?
Trails are easy. The hard part will be comping the fast man into the crowd. Even if you can shoot him on greenscreen you’ll still have to roto all the people and street furniture that will pass in front of him. How well might a difference matte work, shooting him in an empty street, along with a clean street and a full street? Do difference mattes EVER work properly? 🙂