View Full Version : HC vs CCE + QuEnc (New Files online now)
Amnon82
6th February 2005, 14:57
I did a little Quality-Test with CCE 2.70 SP, QuEnc 0.59b2 and HC 0.01
You can see the results here (http://www.brckomania.net/Encodertest/Encodertest.html)
The Settings are:
CCE
Bitrate: Min. 0, Avg. 6000 Max. 8000
Matrix: AutoQ2
Filter: No Filter
GOP: M=3 N/M=4
V/C=30 Pass=2
Quantizer Quantization: 30
Intrablock DC: 9
Progressiv + ZigZag
Aspect: 16:9
HC
Bitrate: Avg. 6000000 Max. 8000000
Matrix: AutoQ2
CPU: SSE2 Profile: BEST
GOP: 12 2
Scene Detection: ON
VBV Check: ON
DC Pred: 9
Scanmethod: ZigZag
Aspect: 16:9
QuEnc
Bitrate: 6000 VBR Max. 8000
Matrix: AutoQ2
High Quality: ON Use Trellis: ON
GOP: 12 2
Scene Detection: ON
2 Pass Enc.: ON
DC: 9
Extrem+Slow Settings: ON
Aspect: 16:9
Fishman0919
6th February 2005, 23:13
To be fair, you should set CCE V/C to 0. 30 is a bit high and limits CCE VBR too much.
Edit - spelling
Rockby
7th February 2005, 00:24
Hi and thanks for the comparison. But..
...i think the bitrate is too high to see any noticeable differences between the Encoders. Set it to an value which is more "realistic". eg. 3000 kbps.
PS: I like the Google-Translator. It's so funny. :)
hank315
7th February 2005, 01:38
Nice work Amnon.
Looked at the movies and they all looked good, I agree with Rockby a lower birate would show the flaws of the encoders in a better way and I mean real low, 1500-2000 Kb/s.
But if you want to do this kind of encodings I really don't see the need to spend a lot of money on CCE, loaded the 3 movies in MPEG Stream and did a frame-frame comparison. The differences were minimal but CCE came out worst IMHO, are you sure you used the best settings? (can not test it myself, don't have CCE)
Some other things I noticed
- Scene change detection in CCE isn't that good, it just misses a lot of them.
- Quenc inserts a lot of I-frames in high action scenes.
But HC did a nice job compared with the "big guys" but maybe I'm not that objective as I should be :D
NOTE1: If you really want to use the best settings for HC, use the autogop option, HC can do that, the other 2 can't :)
NOTE2: Really liked your AUTO Q2 matrix, just set it as one of the standard matrices in the new version of HC; or is it heavily copyrighted :D
NOTE3: Just finished my GUI version but I like your layout more... Will do some more tests and put it online tomorrow.
Amnon82
7th February 2005, 02:05
THX for adding my matrix to HC. Waiting for new HC
dragongodz
7th February 2005, 02:26
ye i have to agree, 6000 is a bit high to really stress any of the encoders(or it should be anyway :) ).
try a series of lower bitrates such as 4000 then 3000 then 2000 and you would be able to see any degradation as the bitrate goes down. just a little bit of work. :)
- Quenc inserts a lot of I-frames in high action scenes.
yes QuEncs scene detection is a bit too sensative for my liking. i have made a few suggestions to Nic already about internal changes for the next QuEnc , may have to mention that aswell.
But HC did a nice job compared with the "big guys" but maybe I'm not that objective as I should be :D
:eek: :D
Really liked your AUTO Q2 matrix, just set it as one of the standard matrices in the new version of HC
quite a little collection of matrices building up in HC i see. ooh look QLB aswell. ;)
hank315
7th February 2005, 03:01
quite a little collection of matrices building up in HC i see. ooh look QLB aswell.;) But ofcourse, it's just one of the best, just wondering who made it :D :D
dragongodz
7th February 2005, 03:13
just wondering who made it :D :D
damn, where did i put my big (C) stamp. ;)
Amnon82
7th February 2005, 03:21
@hank: You like my gui. Thats great! I hope it'll be a inspiration for you ;)
@all: I'll add scenes with lower bitrate, but I don't have the Matrix-Revolutions on my harddrive anymore. I'll look for another good source ;)
... I'll use Blueberry. I encoded with 2000, 3000 and 4000. Screenshots will come later.
Changed Settings:
CCE: V/C 0
Encodertest #2
Here the results of my review:
Bitrate 2000:
Place 1: HCEnc (best Quality)
Place 2: QuEnc (Better than CCE)
Place 3: CCE (worst!)
Bitrate 3000:
Place 1: CCE (best Quality)
Place 2: QuEnc (very close)
Place 3: HCEnc (very close)
Bitrate 4000:
Place 1: HCEnc (best Quality)
Place 2: CCE (very close)
Place 3: QuEnc (very close)
Final Result: (Place1: 3 Points; Place2: 2 Points; Place3: 1 Point)
Place 1: HCEnc with 7 Points
Place 2: CCE with 6 Points
Place 3: QuEnc with 5 Points
Download of the files:
http://www.brckomania.net/Encodertest/Encodertest2.part01.rar
http://www.brckomania.net/Encodertest/Encodertest2.part02.rar
http://www.brckomania.net/Encodertest/Encodertest2.part03.rar
http://www.brckomania.net/Encodertest/Encodertest2.part04.rar
http://www.brckomania.net/Encodertest/Encodertest2.part05.rar
http://www.brckomania.net/Encodertest/Encodertest2.part06.rar
http://www.brckomania.net/Encodertest/Encodertest2.part07.rar
http://www.brckomania.net/Encodertest/Encodertest2.part08.rar
http://www.brckomania.net/Encodertest/Encodertest2.part09.rar
http://www.brckomania.net/Encodertest/Encodertest2.part10.rar
http://www.brckomania.net/Encodertest/Encodertest2.part11.rar
http://www.brckomania.net/Encodertest/Encodertest2.part12.rar
dragongodz
7th February 2005, 16:08
good to see the free encoders are sticking with the commercial ones(ok just 1 in this test :) ).
ok a test one just 1 source is in no way conclusive for all footage types etc but its still interesting. thanks Amnon82.
gigaset2004
7th February 2005, 17:37
@ all,
I did some 1 CD encodings with HC and the result isnīt nice at all.
The sources are Tv material and they are well encoded at 704x576 with Quenc and QLB.
I have also tested some diff.matrices and the result are nearly the same depends a little on the movie.
at least, Hc is too slow at the moment and I hope for further improovement.
canīt wait :-)
so hurry up hank !
Amnon82
7th February 2005, 19:58
I used a source which has all scenes in it. (So think I.)
Hope You like the results ;)
freelock7
7th February 2005, 23:56
Hard to make comparison between these encoders which have their owns standard settings.
CCE is a little bit noisy. So, adding a low pass filter=7 and a vertical filter=4 is a good correction to look like QuEnc soft picture.
HC is the low bitrate winner (no blocs). I agree.
Amnon82
8th February 2005, 00:01
I didn't want to use filters in CCE. I thougth this will give me almost the same picture like the original. The noise I noticed also. HC was the noiseless encode in my eyes. Low bitrates is the biggest problem with CCE...
dragongodz
8th February 2005, 01:49
I used a source which has all scenes in it. (So think I.)
you would need to do sources that were noisy/grainy, clean, very dark, very bright/colourful, high action, slow/static, real life, animation, cg and all the combinations of all those together, eg. grainy+dark+action+life ,grainy+bright+action+life, grainy+dark+action+animation ,grainy+dark+slow+life etc etc etc. to cover every possible type of footage/movie. oh and of course you would want to do them all at the different bitrates aswell. :)
got a couple of spare weeks to spend on just encoding non-stop ? ;)
your tests have ,however, started to show that free encoders are capable of competing with a commercial encoder. so again thanks.
now anyone else want to add some more tests to pinpoint areas that need more work ? :D
freelock7
8th February 2005, 08:39
I didn't want to use filters in CCE. I thougth this will give me almost the same picture like the original
You can increase the low pass value and vertical filter value to reduce noise but you loose in definition. You have to see it!
The noise I noticed also. HC was the noiseless encode in my eyes. Low bitrates is the biggest problem with CCE...
No less noise with HD compared to CCE. They work really closer-without filters.
For QuENc, it's different. The libavcodec settings inside (spacial-temporal masking?) produces a beautiful result. For me it is the winner for its capability to reduce noise from the source.
The others don't do that.
Amnon82
8th February 2005, 14:14
Everybody is welcome to do tests. For me the results so far are, that the freeware-encoder get better and better. TMPGEnc starts also as a freeware software. Lets see what HCEnc will be at the end ...
onesoul
12th March 2005, 03:50
Originally posted by hank315
...loaded the 3 movies in MPEG Stream and did a frame-frame comparison.
What is the MPEG Stream you are talking about here? Can it also load avs files?
(Congratulations on your HC encoder btw)
edit: what does V/C refers to?
Thanks
ReinerSchweinlin
12th March 2005, 20:58
I justr did a quick-test with cce 2.67, tmpegenc.plus, hc0.13 an quenc 0.59 beta. took the first 10000 frames of a stargate.atlantis dvd and just encoded everything as it ist, without filtering, cropping or resizing. bitrate is 1mit/s (heavy low..), PAL source, 2 pass where avayilabel, 9bit DC precision and most precise/slow settings on all encoders. CCE did 3 pass. progressive flags were set
Results are astounishing!!!
Both commercial Encoders looose so clearly!!! Tmpgenc still is quite watchable, cce isnīt even worth looking at. I have to review my settings for cce, i can īt belive the result.
Quenc did the best job, most scenes are very good. Only movement reveals the low bitrate, but it still is watchable. hc closes up quite good, folowed by tmpgenc, which does a great job on stills but introduces heavy blocking on movement. cce almost looks as if there were only i-frames with a quant of 50 !!!
SiXXGuNNZ
12th March 2005, 23:58
Encoders Used:
CCE SP v2.70.02.00
HC v0.12 beta
QuEnc v0.59 [BETA 2]
Bitrates encoded to: 2500, 3500, 5000
Matrix used: MPEG Standard
8, 16, 19, 22, 26, 27, 29, 34
16, 16, 22, 24, 27, 29, 34, 37
19, 22, 26, 27, 29, 34, 34, 38
22, 22, 26, 27, 29, 34, 37, 40
22, 26, 27, 29, 32, 35, 40, 48
26, 27, 29, 32, 35, 40, 48, 58
26, 27, 29, 34, 38, 46, 56, 69
27, 29, 35, 38, 46, 56, 69, 83
16, 17, 18, 19, 20, 21, 22, 23
17, 18, 19, 20, 21, 22, 23, 24
18, 19, 20, 21, 22, 23, 24, 25
19, 20, 21, 22, 23, 24, 26, 27
20, 21, 22, 23, 25, 26, 27, 28
21, 22, 23, 24, 26, 27, 28, 30
22, 23, 24, 26, 27, 28, 30, 31
23, 24, 25, 27, 28, 30, 31, 33
I chose the standard matrix because this was testing the general quality at certain bitrates for these encoders, not a matrix comparison.
as for settings, all gops were 12/2, all the best settings were turned on for each encoder, if you would like the settings, ask and I will post them.
all encodes were 2 passes.
so here are the compare files:
http://tehru.org/sxgz/mpeg2_compare/
Ebobtron
13th March 2005, 03:44
I would like to see some PSNR charts over about 1000 frames. My eyes are not at all a good measure. I can compile the numbers and gen the charts, but I do not have the fancy encoders. QuEnc and HC really come out close at the higher bit rates of 5000-6000.
------------
Anybody want to put our efforts together we could measure what we cannot see.
zilexa
20th May 2005, 00:09
I am using DIKO to create 3 DVDs with each 8 episodes of Lost (source is XVID, 350 MB, 42 minutes of video, audio is MP3 VBR between 110-175 kbps)
Now this means the bitrate for MPEG2 will be low (between 1.9 and 2.9).
Wich encoder would be recommended?
Or, perhaps a better question, wich one would not be recommended?
I think the battle goes between FreeEnc (same as Quenc but newer and optimized?) and HC.
I used CCE SP 2.67... 4 passes. quality was pretty good, not as good as original but still very good! But it took me 27 hours (Pentium-M 1.6 Ghz).
So I would like to know wether to use FreeEnc (0.44 now) or HC..
Archimedes
20th May 2005, 00:48
Can you give me a link to the source video of the test above.
Thanks in advance
Josef
dragongodz
20th May 2005, 03:31
FreeEnc (same as Quenc but newer and optimized?)
if you read in the QuEnc threads there have been updates etc so newer isnt really right.
Freeenc does now use a new/modded version of rate conrol which should be better than the normal libavcodecs. however QuEnc is also coming soon with a new rate control, not the same as Freeencs, so you may also want to give that a try when its released.
interesting times for free encoders. :D
Encoder Master
21st May 2005, 09:12
however QuEnc is also coming soon with a new rate control
This would be great because I like the quality of QuEnc. This avcodec based front end gives me the best quality of this types of encoder (FreeEnc, NuEnc, etc.)
TomBrooklyn
15th November 2005, 22:04
ye i have to agree, 6000 is a bit high to really stress any of the encoders.
Hi,
I am using QuEnc with RB. How can I tell know what bitrate it's set at?
And what bitrate setting is best to get the best quality? I don't mind encoding taking a long time as I run RB when I go to sleep or am at work.
DK
15th November 2005, 23:38
in rb you could click on MODE => 1click (in case 1click was ENabled before)
clicking this entry will switch to 3 click method
NOW you can use PREPARE and a couple of seconds later rebuilder will state the bitrate like this:
- HIGH/LOW/TYPICAL Bitrates: 3.993/600/3.178 Kbs
TomBrooklyn
16th November 2005, 01:16
Hi DK,
I did as you suggested and I got HIGH/LOW/TYPICAL Bitrates: 4,321/501/2900
Are these adjustable somehow, and if so, how? If not, how are the bitrates set?
DK
16th November 2005, 01:26
adjustable?
not quite
you would have to alter your source - i.e. removing parts you wouldnt want
that of course would increase overall bitrate
another way could be reducing extra/bonus material to half d1 resolution and *stelaing space from extras by n%*
this, too, will increase the bitrate for the main movie slightly
on the other hand why dont you give it a try with the current numbers?
~3000kbps isnt that bad
TomBrooklyn
16th November 2005, 01:42
Hi DK,
OK, I think I see, the bigger the source to encode, the lower the bitrate has to be to be able to fit it into the allowable space.
Is the last number, the "Typical" number, the one that is usually being referred to when talking about "the bitrate"?
What bitrate do most people consider to provide an excellent picture, a good picture, a decent picture, etc?
DK
16th November 2005, 02:06
not the BIGGER the source but the LONGER the playing time
the last value is the average bitrate in order to achieve the calculated/desired final size - i.e. ~4470mb for your final dvd output
well judging only by the numbers isnt useful
2000kbps can be sufficient with low action movies
3000kbps are often considered the starting point for decent quality
vBulletin® v3.8.5, Copyright ©2000-2012, Jelsoft Enterprises Ltd.