Andrew Somers
Forum Replies Created
-
8 cores is fine – but 8 GB of RAM is NOWHERE NEAR ENOUGH.
For an 8 core machine you want 24 GB MINIMUM. (a good rule of thumb is 4 GB PER CORE minimum). However, the Westmeres also have the advantage of “virtual cores” (2 threads per core) – depending on software, you may then actually want 48 GB.
But 24GB is a minimum for 8 logical cores in any case, assuming you want to use multiprocessing to good advantage when rendering.
AE and Premiere share libraries, so you can run both without a “substantial” increase in your RAM footprint.
Video Card:
Regarding the suggestion of the nVidia card: At the moment only Premiere takes advantage of CUDA, After Effects does not – YET – However in CS6, AE will be CUDA accelerated.The CUDA enabled nVidia Quadro 4800 card is VERY expensive relative to the ATI 5770 – system RAM is more important. Adding a $1500 nVidia card and then NOT buying a few hundred dollars worth of RAM would make no sense. Buy the RAM first,
If ou are using Premeire for EDITING, then you would probably prefer to have a proper SDI/HDMI card such as the BlackMagic Decklink.
the entry level SDI decklink is only $300, and DOES offer real-time effects for Premeire, etc. and support of the Mercury Playback Engine
https://www.blackmagic-design.com/products/decklink/
-
he obscuration layer is more of a convenience – but is not the best solution in all cases.
In YOUR case, use TWO separate layers of Particular, one for front and one for back.
-
I highly suggest Nikon. I wrote a few detailed posts just a week or two ago here:
Regarding the Nikon D7000 (by far the best video-capable dSLR in the under $3000 class, and at $1300, you won;t find a more capable camera):
https://forums.creativecow.net/readpost/280/8285
And regarding a few useful lens choices for the D7000:
https://forums.creativecow.net/readpost/280/8316
If you have more to spend, the Nikon D4 is the best video camera below the Red Scarlet or Epic.
The D4 has superior dynamic range and high ISO performance compared to any camera (still or video) under $10,000. The D4 Body is $6000.
https://www.nikonusa.com/Nikon-Products/Product/Digital-SLR-Cameras/25482/D4.html
At a more modest $3000, the D800 utterly destroys the Canon 5D and 7D in terms of performance for both video and stills:
https://www.nikonusa.com/Nikon-Products/Product/Digital-SLR-Cameras/25480/D800.html
See this comparative review at DXOmark, comparing the Nikon D800 vs the Canon 5D Mark II:
Now, as to the Sony FS100 – there is a lot to be said for a dedicated camcorder instead of a dSLR. For $5000 for the body, it looks to be a solid camera – though the D4 exceeds it in terms of ISO and dynamic range.
Both the D4 and the FS100 have full res HDMI outputs (so you don’t have to use the built in compression). But both are also 8 bit in terms of video.
Now there is talk that the FS100 will have an upgrade to 10 bit – and THAT would really swing in favor of the FS100. A true 10 bit camera for $5000 ? That would be a real winner.
On the other hand, THIS GUY seems to hate it:
https://wideopencamera.com/skid-row/10-reasons-to-hate-the-sony-fs100/
-
Yes, AfterEffect’s tracker gets very slow on large tracking selections.
Also, AE’s tracker likes pointy edges better than round object IMO. I thin you’d do better to track the small dirt/marks around the perimeter than the large round silver object. If you track several constrained points, you can use an expression to average them out for a more stable track.
HOWEVER,
I stopped using AE’s tracker completely once I started using Mocha. (or the Pixel Farm products for that matter).
-
I need more information.
*What kind of TIFF sequence (what resolution and bit depth?)
What in the image are you tracking? Please provide a screenshot of your tracking marker placement and search area size in the frame, and an example of the full frame.
-
Green Screen is the color of choice for most applications where a color-type key is needed particularly when shooting with a digital camera.
For one, Digital cameras with bayer sensors have better resolution in green (there are twice as many green pixels as red or blue at the sensor level), and more importantly, the green pixels are of a much lower noise than the blue.
But to be successful, the green used should be of a sufficiently narrow band that there are several stops difference between the green and the red/blue channel exposure.
On the other hand, the green makeup on the witch’s face is probably not going to be so narrow-band. Provided there is enough image data in the red (or blue) channel, you may still be able to get a good clean key.
BUT DO TESTS.
I can not stress this enough. Make some tests with your intended camera, under intended lighting and against the specific green color intended for the screen, and a stand-in with the makeup on. Then load that footage into after effects, and examine the color of the screen using the eyedropper tool.
In a 32 bit linear colorspace, You should see values like this on the greenscreen portion of the image:
Note the very large differences in value between the red and blue channels. You need this kind of channel separation for the green screen.
After you do tests on the witch makeup, you may find that you need to make no real changes. Or you may find that you need to adjust it.
My suggestion is to make the witch makeup green but with a substantial red component. Then later in color correction you can do a secondary color correction on that specific color, and rotate the hue back toward green.
In the below example (not a real world test obviously), a small shift in the tone of the makeup was enough to keep the face out of the “green screen” and yet was very easy to selectively adjust the face color with nothing more than a 45 degree rotation of HUE, selectively applied only to the yellow/green color.
NOTE: For some reason, Creative Cow is altering the color green in this image when uploaded here, so this may not look like the correct green on your monitor. This despite the fact that the image is tagged with am sRGB profile – it seems that this website does some transform and ignores the tag.
-
If your export looks different than your preview, it is usually to do with OpenGL.
As Dave mentioned you should not use OpenGL to render anything intended for “final use”, but I’d like to add that even if you have OpenGL turned OFF in rendering, if you are using OpenGL to preview, you will still have a discrepancy between what you see on screen, and what comes out of the render engine.
Prior to render, it is a good idea to look over your comp with “Fast Previews” turned off before render.
Alternately, in my project template, I have a “work comp” where I do all my work, and that is nested in a “render comp” that contains the slate and other things – the render comp is setup with no fast previews, conversions to final color space, what have you – and I will look at the render comp for making final adjustments for color, etc.
-
This probably should be in the “Expression” forum.
I just pasted your code into the Y rotation of a layer, I I think it is working as you would want it to.
How are you controlling the white level? I’m using levels…
-
In expressions you can specify time relative to what you are evaluating.
You are looking for the “valueAtTime” method. Here’s an example:
https://help.adobe.com/en_US/aftereffects/cs/using/WS3878526689cb91655866c1103906c6dea-7a15a.html
valueAtTime is discussed in the property attributes reference:
https://help.adobe.com/en_US/aftereffects/cs/using/WS3878526689cb91655866c1103906c6dea-79f8a.html
Also you may want to use time conversions for some items:
https://help.adobe.com/en_US/aftereffects/cs/using/WS3878526689cb91655866c1103906c6dea-7a09a.html
-
…Which would be useful if I watched madonna video clips…
You are coming to this free forum to ask a question, it’s not too much to ask that you post a link and or image that is quite specific of what it is you are trying to do. If there is a specific video, l=then link to it and state the specific time in the video that the effect occurs.
