Hi all,
I have been trying to make my first KDVD using vmesquita's guide and TMPGENC 2.5. I've made simple KVCDs before so I understand some (although not all) of the video principles, but I am really stuck since my calculations using this method are not working as expected from the guide. I've spent over 3 hours looking through this forum and others looking for answers but I can't find anything that seems to solve my problem, which is why I'm writing. I've tried to give all my calculations below (hence the length of the post!) in the hope that someone can see the flaw.
I don't have access to CCE (requires XP if I remember correctly, which I don't have - also why I am not using DIKO - any chance of a non-XP version of this vmesquita? I'd be willing to re-compile the source for you if needed on my ME box - and have Visual Studio to boot (software engineer by trade) ) and have been trying to follow the TMPGENC part of the guide but I am not getting anywhere (certainly step 11 for TMPGENC is rather ambiguous with regards to which quality method to use (CQ? CQ_VBR? VBR? etc.) ).
Maybe it's my calculations, maybe it's TMPGENC, maybe it's the guide with respect to TMPGENC - not really sure to be honest, and I could use some help to figure out where I am going wrong (cause something has to be wrong somewhere!!
)
I am trying to put 8 episodes of a TV series on a KDVD. All source is originally 512 x 384 (except last which is 576 x 320) and all sources are PAL (25.0 fps, GOP = 15). Don't know whether PAL is causing a difference to calcs but...
The episodes timing and frame rates are:
Frames Time
Ep 1 73423 0:48:56.92
Ep 2 73436 0:48:57.24
Ep 3 73901 0:49:16.04
Ep 4 73573 0:49:02.92
Ep 5 73923 0:49:16.92
Ep 6 73840 0:49:13.60
Ep 7 73349 0:48:52.92
Docum. 54371 0:18:37.08
TOTAL 543371 6:02:10.484 (~362 min)
Therefore, since a DVD has 4282 Mb, and every minute of 192 kBps audio is ~1.4Mb, then the audio on the DVD should be 506.8 Mb, leaving 3775.2 Mb for video (4282 - 506.
Therefore, using your calculation for the "sample factor" which I will call x (anyone have a better name for this?), I get
x = (543371 / 362) / 15
x = 100.0683
Therefore, the sample length (L) should be
L = 3775.2 / 100.0683
L = 37.73 Mb
Now this is for the ENTIRE portion of video. I am guessing (and this could be a source of error) that to get the sample size for any given episode, I should multiply "L" by the percentage of time an episode takes of the entire video - does anyone disagree?
So for Episode 1, the sample factor should be
L = (49 / 362) * 37.73 Mb = 5.107 Mb
I then have an AVS script for the video portion of the episode, which is as follows:
Code:
========= Script Starts ===========
## PLUGINS
LoadPlugin ("D:\Video\AviSynth\Filters\blockbuster.dll")
LoadPlugin ("D:\Video\AviSynth\Filters\atc.dll")
LoadPlugin ("D:\Video\AviSynth\Filters\grip.dll")
LoadPlugin ("D:\Video\AviSynth\Filters\sampler-2.5.dll")
######################################################################
# Based on Optimized Script to remove DCT Blocks by Girv and DialHot #
######################################################################
##############
# Parameters #
##############
#Use 720 or 352 to KDVD, 528 352 to K(S)VCD
m_width=720
myoverscan=0 # Test values from 0 to 3 here. Bigger files produce more quality but
#a value too big will cause undesirable black borders on your TV.
##########
# script #
##########
AVISource("ep101.avi",false)
#Post-Process
Blockbuster(method="noise",detail_min=1,detail_max=8,variance=0.3,seed=5823)
ATC(1,2,4,0.5,false)
TemporalSoften(2,7,7,3,2)
Blockbuster(method="noise",detail_min=1,detail_max=10,variance=0.5,seed=5823)
#Resize, overscan and borders
GripCrop(m_width, 576, overscan=myoverscan,source_anamorphic=false,dest_anamorphic=false)
GripSize(resizer="lanczosresize")
GripBorders()
# Get sample length
sampler (length=15)
========= Script Ends ===========
I load this AVS script into TMPGENC 2.5 Plus, using the PAL KDVD Full Template (KDVD-720x576-Full-D1-_PAL_.mcf). I use the Project Wizard to load it in and it claims that it needs to judge the field order and says that it's Interlaced, Bottom Field First (field B) and 4:3 625 line PAL (704 x 576). I don't do anything else in the wizard except set the size to 720 x 576 in the last step of the wizard.
The settings defined by TMPGEnc include:
Video Tab
---------
Aspect Ratio 16:9
Frame Rate: 25 fps
CQ: 70 (min 300; max 5000; enable padding not be lower than min frame rate; P & B spoliage both 0)
VBV Buffer Size: 224 Kb
Profile & Level: Main Profile & Main Level (MP @ ML)
Encode: Interlaced
YUV: 4:2:0
DC Component Precision: 10 bit
Motion Search Precision: Highest Quality
Advanced Tab
------------
Src / Field Order / Source Aspect: as in Project Wizard
Video Arrange: Full Screen
The problem is that when I set a CQ between 60 and 70, the sample size ranges between 7.5Mb and 10Mb,
NOT 5.107Mb as calculated by the giude. I know from experience that I don't want to get the CQ below 60, but I am at a loss what is going wrong? Should I be using a different quality type such as VBR? Change the default CQ range (e.g. change 5000 to 6000)? A differnt TMPGENC setting? Is something wrong in my calculations? Or is something wrong in the TMPGENC portion of the guide? I really don't know and am all confused - help!!
So if anyone could let me know what I am doing wrong, I'd appreciate it since I can't see the flaw myself
Thanks again for reading this!
John Lawrence
PS to vmesquita: A suggestion for next version of your (excellent) doc - an actual worked out real-life test case would be nice to have - so one could see where all the calcuations go, etc. - perhaps use my data for it?? You're certainly welcome to it!
