Kevin Camp
Forum Replies Created
-
opengl… now we’re getting somewhere.
i’ve seen a strange issue with opengl on a macpro with ae7, where if you apply an opengl accelerated effect on a layer (with opengl enabled) ae crashes, but if you apply it to and adjustment layer it works great. i realize this doesn’t relate to you problem, but it does show that opengl in ae can do strange things.
what g5 (dual, quad? mhz?) and graphics card do you have?
i’ll have to double check, but i believe that opengl acceleration in 6.5 did not support motion blur (as an accelerated feature). and i’m thinking that you may have a graphics card that is opengl 2.0 (which does support accelerated motion blur) but that may be creating some sort of issue with ae6.5.
whether this is the case or not, i would try disabling opengl for previews in the preferences. then check a ram preview for motion blur. if motion blur works without opengl on then the final render should also work (they will be using the same software render engine with opengl disabled).
if the ram preview has no motion blur with opengl disabled, re-inable it and set the render settings to use opengl for final renders (i’m pretty sure 6.5 has a check box in the render settings). see if the render has motion blur.
Kevin Camp
Designer – KCPQ, KMYQ & KRCW -
years???!!!
well, you’ve probably checked this already, but there is an update for 6.5 available (6.5.1).
have you tried posting in adobe’s ae forum? how about re-installing ae? or, download the beta for cs3 and see if mo-blur works in that version, if it does… upgrade next month.
maybe you could get it to crash… i’m much better at troubleshooting macs that crash, i’ve never heard of motion blur not working.
you didn’t mention it, but i assume opengl is not enabled for renders (in the render settings) or previews (in the preview pref).
Kevin Camp
Designer – KCPQ, KMYQ & KRCW -
double check the best settings, make sure that motion blur is on for checked layers. this shouldn’t be a codec issue…
are you using opengl for previews? i’m kind of grasping here, but i’m looking for differences in preview settings versus render settings.
Kevin Camp
Designer – KCPQ, KMYQ & KRCW -
how about a freeze frame…
too fake…? how about a freeze frame of the phone, cut out in photoshop. comp that with the text, the comp that over a background shot with some movement… you could even add some movement back onto the phone-with-text comp.
Kevin Camp
Designer – KCPQ, KMYQ & KRCW -
-
the yuv to rgb conversion is the gotcha here. it seems to be impossible to go from yuv to rgb and back to yuv and have it look exactly the same.
what seems to be your best bet is to try and export your footage from your nle with a conversion from yuv to rgb to a lossless or uncompressed format. take that into ae, render back to the same format. import that into your nle converting the rgb to yuv and see how close you are. (unfortuanately, the nle will recompress the lossless/uncompresseed footage back to the compression used to capture the original, so some quality loss will occur).
you can also compare that to using ae to do the yuv conversion. take your yuv footage, choose interpret footage from the file menu, click the check box for expand itu-r 601 luma values. drop that into a comp, then add a levels adjustment with black output set to 16 and white output set to 235. render and import into you nle (no conversion for color space) and compare.
choose which looks best to you, and use that workflow, but i think you’ll always get some shift and slight quality loss.
Kevin Camp
Designer – KCPQ, KMYQ & KRCW -
well if video for windows works for you, i don’t think there is any problem with it (i actually don’t have that setting, so i don’t know exactly what it’s doing). i can’t think of any reason why the lossless animation codec would change the quality of the output, so i still think there may be something else happening there…. maybe the animation setting was lower than ‘best’ (there is a slider that goes from least to best, but i’ve never changed it from best).
i have heard that magic bullet works very well, but i’ve never used it. if you have dv footage and want to lessen the artifacts a little, ae7 has an effect preset called reduce dv blockiness that can help. if you are keying dv there are a few article/turorials here about keying dv, search the articles for keying and dv (or just keying for other helpful techniques); aharon robinowitz, andrew kramer and barend onneweer have some for ae specificly. just scroll down those pages to see their turotial libraries.
in addition, ae also has some noise or grain removal effects that may help.
personally, i wouldn’t try to go 24p at this point. it will be difficult to recreate the natural motion blur that you would see if the footage was shot at 24p, so it will look a bit jerky or strobe-like.
if you wanted to try it, the smoothest way i’ve found is to make sure the footage is interpretted correctly (selct the footage in the project window, then choose file>interpret footage>main) make sure the separate fields in set to correctly (lower for ntsc dv), you may want preserve edges checked (check it out both ways). then drop it into a 24p (23.976fps) comp and enable frame blending for the comp (the film strip icon) and turn it on in the timeline for your footage (click in the empty box below the film icon 2 times, so it looks like a smooth diagonal line).
you may be able to help smooth the motion by keyframing a little directional blur or with an effect by re:vision called reelsmart motion blur, but i would just wait for your next project and shoot 24p at that point (i’m not familiar with your camera, but many can do this). then you could also look into creating a progressive dvd in encore to avoid having to add the pulldown.
Kevin Camp
Designer – KCPQ, KMYQ & KRCW -
i think this just comes down to how the producer/director wants this to look in the end.
if they want it to look progressive in the end, then they should shoot prgressive, the image quality, motion blur and overall look will be noticably better than to de-interlace in post.
if they want it to look smoother, like video, then they should shoot it interlaced and keep it that way throughout. it forces you to deal with removing fields and putting them back in, but that’s not usually too bad of a workflow.
but to go out of their way shoot interlaced just to convert it to progressive in the end is just silly. (unless they wanted to slow-mo the footage, which actually can produce pretty good results)
Kevin Camp
Designer – KCPQ, KMYQ & KRCW -
if you can capture the hd uncompressed or straight to the image sequence, your quality should be very good. you will, however, need a fast drive array to capture 1080 uncompressed and a capture device that supports uncompressed capture.
is the 1080 d5 tape interlaced (24p with a pulldown)? if so, you would like to remove the pulldown before creating the image sequence. you may be able to do this at capture or you may have to do it in another piece of software (ae can do this easily).
if you are having someone else do the capture, make sure they can capture hd uncompressed. and, if your tape has a pulldown added, ask if they can remove that too.
Kevin Camp
Designer – KCPQ, KMYQ & KRCW -
have you looked at cc rain and cc drizzle in the simulation effects?
if you don’t see them, look on your ae install disc for a folder called cycore fx in the third party effects folder and run the installer. they are free, fully functional effects.
you might also want to install keylight and color finess while your at it.
Kevin Camp
Designer – KCPQ, KMYQ & KRCW