I have posted on this subject before. I am capturing Dishnetwork with a Canopus ADVC-100. I am then compressing with TMPGEnc to a 250 GB hard drive, not to DVD. I want the smallest file size with no quality loss.
I am using a resolution of 704 x 480. I am using the Constant Quality (CQ) mode with Quality set at 65.
I was using a max bitrate of 3400 but this seems to result in quality loss in terms of resolution/sharpness. I increased this to 5000 and it seems better.
How does one tell the Max bitrate to use without losing quality and keeping the file size as small as possible? It is hard to do subjectively.
Thanks much!
+ Reply to Thread
Results 1 to 10 of 10
-
-
Constant quality at such a low bitrate is your problem here... It just won't look any good at that rate.
Switch to 2-pass VBR with an average of 2500, a max of 3500, a min of 0 and you should see a marked improvement in quality with a similarly smaller file. You could even try dropping the average down to 2000 and see if it is acceptable.
Take a small capture that you have (or make one to test with) and then try it out, I think you will find the difference worth the extra encoding time.
Regards,
Savant -
Originally Posted by neo
There simply is no definate answer to this question for several reasons
Quality is simply too subjective.
Different types of material requires different bitrate settings for the same quality (Studio discussions and talking heads can get away with low bitrates where fast action movies or music videos with lots of dancing and flashing lights need higher bitrates).
Try switching to encoding at 352*480 and see if that quality is acceptable to you at the lower bitrates you desire. -
Originally Posted by Savant
When I was testing 2-pass it seemed to give worse quality at the same file size compared to CQ. This was some time ago but as I recall from studying the log files the first pass did not appear to take samples often enough to represent the rapidily changing bitrate (BR) needs. It also appeared that in the second pass there was a much lower difference in the variation of bitrate. In other words, when the material did not need much bitrate the CQ dropped the BR much lower and then increased it much higher. It seemed much more responsive for each GOP.
With CQ at 65 and max BR at 5000 this looks much better. I believe the 65 sets the average BR at 65% of max but I am not sure on this. If so, this should give me an avg BR of 3250.
At these settings I get about 85 minutes/GB.
neo -
I think your bitrate is too small for your frame size. At this low bitrate, the pictrue is going to look a bit the worse for wear. For bitrates under 4Mbps I would only use Half D1 frame sizes (ie. 352x576 (480)). If you want full (or Cropped) D1 then you need to start looking at bitrates of 6Mbps plus. With TMPGenc, CQ coding is much better quality than multipass VBR, but the file size will suffer. To get decent results, I would recommend using Half D1 res.
-
Originally Posted by energy80s
With my testing, I found CQ to give better quality at the same file size due to the greater dynamic range in BR per GOP with CQ as can be seen in the log files.
Are you suggesting I move the max BR to 6 with a Quality of 65?
Thanks much -
I normally do full D1 encodes with the CQ set to 80 and max bitrate at 8000 and min set at 4000. This will make a large file however. I imagine that Dish Network is something similar to Sky Digital here (shite quality) but to re-encode this material needs a higher bitrate than was used originally by the broadcaster. So even though DISH may only be using bitrates of 4Mbps, you will need to increase this "just to stand still" and not drop off the picture quality. I suppose 6 and 3 could be used as the max and min settings, with a CQ of say 70 or 75. With TMPG it is important to have the different bitrate settings as "half" or strange things can happen. This is not true of other MPEG encoders.
-
Originally Posted by energy80s
neo -
I decided to look at the help file in the program and it explains more. It explains the modes. In CQ mode the Quality controls the B frame spoilage, not the average BR compared to Max. I am thinking that it uses whatever BR it needs at the time, not to exceed the Max. This makes more sense because when I increased the Max BR from 3400 to 5000 there was little increase in file size.
I use a GOP structure of 1, 5, 2: IBBPBBPBBPBBPBBPBB (18). Since 2/3 of my frames are B frames I tested changing the Quality. When I changed the Quality from 65 to 100 the file size for a short 4+ minute test increased from 70 MB to 226 MB. I did not see that much improvement in picture quality.
Now that I understand this better, I will do some more testing both with BR and Quality.
neo
Similar Threads
-
A ffmpegX's problem about setting the audio bitrate
By lzero1319 in forum ffmpegX general discussionReplies: 1Last Post: 7th Apr 2012, 08:44 -
Jittery movement on encoded file (Bitrate setting question)
By cb678 in forum ffmpegX general discussionReplies: 4Last Post: 5th Nov 2011, 14:30 -
MKV to AVI Video Bitrate Setting
By gonwk in forum Newbie / General discussionsReplies: 6Last Post: 8th Apr 2011, 19:31 -
Xvid Setting: 1pass Quantizer vs Bitrate Mode, vs 2pass => Quality
By carlaron in forum Video ConversionReplies: 12Last Post: 4th Oct 2010, 07:12 -
TMPGEnc 4 Divx Setting
By Anakin in forum Video ConversionReplies: 2Last Post: 14th Oct 2009, 16:21