How important 1 pass VBR or 4 passes?

[BrainSoup]

Is it really do any difference when u do 4 passes VBR or 1 pass VBR when converting from XviD avi to SVCD?

[DJRumpy]

If your souce is complex (lots of scene changes, fast action, movement), then yes. More passes are ALWAYS better. Think of it as having 15 minutes to get through a complex maze. The multipass essentially hands the MPEG encoder a MAP of the maze, letting it perform better in the time alloted.

Skip this if you don't want the detailed description...

Each additional pass allows your encoder to better allocate bitrate to area's that need it, borrowing bitrate from those frames that don't need additional bitrate. It can do this because the additional pass gives it foreknowledge of what each frame requires. It can give a frame at the beginning of a movie, bitrate from the end frame of the movie, if it knows beforehand that the end frame doesn't need it. It can't do that with a single pass. It also allows the encoder to insert I-Frames when needed to stop complex scenes from exceeding the available bitrate. The farther away from an I-Frame your encoder gets, the more bitrate it uses (assuming some motion in each consecutive frame). As new motion is detected as 'new', bitrate is applied to reproduce those 'new' areas. If enough motion is detected as 'new', rather than just an old object moving in the frame, eventually you end up encoding the entire frame as 'new'. Enough of these 'new' frames in a row will quickly exceed the available bitrate. When that happens, the overall quality is reduced to compensate, and you get macroblocking.

Multipass helps head that off by inserting I-Frames when needed. These reference frames give the encoder a fresh picture of what's new and what's not, so it doesn't wastefully throw bitrate at an all-new frame.

[Poplar]

Multipass is always better than one-pass. But by how much? I did some test somewhile ago. The clip was encoded with CCE One-pass VBR and 3-pass VBR with same bitrates. Visually, I couldn't see the quality difference. When analyzed by Mosalina, 3-pass was only 0.36 point better than 1-pass on a 1-40 scale. Was it worth it? You decide. I decided to use maximum bitrate with one-pass, because I have a slow computer.

[andkiich]

A lot of these differences you'll notice when you migrate fro ma smaller screen to a larger one and go from a tube to a projection.

[Lelon]

Personally, I notice a lot of difference between 1 pass and 2 pass VBR, but little difference between 2 pass and 3 or 4 pass.

[DJRumpy]

The places to look are specifically on scene fades, changes, and high motion scenes. You won't notice on a slow scene because it hasn't run out of bitrate. Use the slow motion mode if your player supports it, or even go frame by frame. Look at the frames directly after a scene change or fade, and examine the whole sequence during a high motion scene. You should see a drop in quality, indicated by macroblocking, blurryness, or any other artifacts. A few good movies to do this experiment on (make sure you try it on SVCD. DVD rarely experiences this due to it's large bitrange and high max):

S.I.G.N.S - beginning of the movie where they are running through the cornfield.
Star Wars - Attack of the Clones - towards the end of the movie, where all of the jedi are surrounded and deflecting laser blasts with their sabers.
Any of the new Marvel Comics movies like Dare Devil - the beginning where the marvel logo pops up.

Try any of those sections, using 1 pass (CBR or VBR), and then try a multipass run, and see the difference. Make sure you do this within standards. It doesn't do much good if your making an xSVCD with a 4000kbps max. You won't see anything.

[spectroelectro]

Assuming your bitrate is set reasonably high (above 5000 or so), the primary difference twixt multipass and 1-pass encoding boils down to filespace. A typical 1-pass VBR encoding at X bitrate using TMPGEnc is roughly 50% larger than size of the typical 2-pass VBR encoding at the same bitrate using the same encoder. (CBR rouhgly doubles the filesize.)
In practical terms, this means that you can cram more video on a DVD-R with visually identical quality if you use 2-pass encoding.
Specific example: I get 6 twenty-five-minute episodes of "Cowboy Bebop" on a single-layer 4.7 gig DVD-R with 2-pass encoding. Using 1-pass VBR encoding I would get only 4 episodes, and with CBR only 3 episodes. Quality looks identical.

[DJRumpy]

Yup. CQ mode gives good quality (assuming your max is high enough), but it's not very efficient on space. Even DVD, with all of it's space almost requires multipass for any total video length over 2 hours, assuming your using full D1, rather than Half D1 or smaller.

[Adder_78]

Try this OPV method:
http://www.mpegit.net/eng_svcd.php

Fast method with same quality as 4-pass multipass vbr with CCE SP. Don't mix OPV and Multipass VBR. Totally two different things...

[FOO]

I would like to know how a video with 5000K average bits per second
can have anything but a fixed size no matter how it's encoded.

If one file is larger than another, then it isn't the same bitrate is it ?

[DJRumpy]

Correct. We're talking changes in qualtiy, not size. CQ mode simply varies the bitrate as needed by the video, but within the constraints set on teh MIN and MAX values. CQ just uses X number of bits per second. VBR allows variance within the values you specify, while meeting X average of bits per second. It allows the encoder to borrow bits here and there, where they aren't need, to be used where they are needed most. The AVG setting in VBR still lets you specify an exact output size, while still being flexible.