Confuse about motion estimate in TMPGENC and ATI AIW Radeon

[tom28]

I am so confuse about the motion estimate setting in TMPGENC and ATI setting. I understand that if you set motion estimate in TMPGENC to good or very good it will give good quality video. In ATI AIW Radeon, the motion estimate on DVD setting is default to zero. In viewing the capture on ATI AIW, with zero motion estimate, VBR or I Frame only, the output is pretty good. So my questions are:

1. What does motion estimate do?
2. If your video doesn't have much of motion, does it mean setting motion estimate to zero in ATI AIW, or fast in TMPGENC will yield the same result (as in good or best in TMPGENC or raise the motion estimate bar in ATI AIW setting)?
3. In I Frame only capture, do you need to set motion estimate higher than Zero? or it applies to VBR, CBR and I frame only.

Thanks.

[Tommyknocker]

Motion search precision: Motion search precision adjusts how much the encoder searches for motion between different frames. This means it looks at a block in a frame and then looks at the next frame for the same (or similar, depending on other settings) block somewhere else in the picture. If a similarity is found, information on the direction and how much the mpeg block moved between the 2 frames is put into the mpeg stream, so that the block is used
in a different place in frame 2. By doing this the same thing isn't encoded twice. This means that less bits are used when things are repeated. then bits can go to other things like encoding the block better in the 1st frame.

Setting the motion search precision to the highest quality (slowest) setting will make the encoder search really hard for repetition between frames. This makes the picture a lot better when things are moving.

If you encode a video with only small amounts of movement, you can sometimes get away with lower settings to save time, but even in low motion pictures, different parts of the image might be the
same, so I always use the highest setting to give the best picture.

Professional DVD authoring hardware uses extremely sophisticated methods to search for motion, and that's what makes them so expensive, but so high quality. - mikk

http://www.vcdhelp.com/tmpgencexplained.htm

[patrickm]

you've got apples and oranges.

in ATI, there is a way to change the maximum size of a motion vector through the registry, so that is similar to what TMPG is doing.

what "Motion Estimation" in ATI really means is "max % of CPU dedicated to encode". it's just CPU utilization. keep it as high as you can, but don't exceed like 97%. that way, you use all the CPU you can, but a minor thing in the background using a few CPU cycles (i.e. <3%) won't cause you to drop frames.