Log in

View Full Version : Which prefered higher resolution for encode to see on Home player ?


Dams
15th November 2006, 11:46
All is in title ...

Let explain your choice and also, let's say what is the size approx. of these HD like encode ..

Dams
15th November 2006, 11:54
For myself, I prefer 688*384 which is the best "quality/price" : good resolution and not to big for frame size

raeltheimperialaerosolkid
15th November 2006, 12:06
I would stick on 704*384 since it's still in the range of my "normal" Xvid stand alone player.

celtic_druid
15th November 2006, 13:00
I'd encode with non square pixels. That would make it around 720x544.

Dams
15th November 2006, 13:47
I'd encode with non square pixels. That would make it around 720x544.

Well , it is for 4:3 format display, or you make anamorphic encode ?

celtic_druid
15th November 2006, 14:18
Anamorphic.

Sagittaire
15th November 2006, 14:30
720*576@720*400 is always an underresize even if you make a large crop 704*560@720*400.

720*400 is highest possible resize (SAP compliance) in the no anamorphic way if you want the best possible quality at high/medium bitrate (less than average q4 for H263 quantisation with XviD/DivX). IMO actually average q4 720*400 done better visual result than average q3 640*352.

Dams
15th November 2006, 14:51
Yeah, I found when I change from my old home player to my new Sylvercrest 6515, the picture was better but I feel that little resolution (like 640*272 or 608*336 for example) wasn't so detailled.

So I decided to "push" higher the resolution of my encode.

Well, what target do you usualy achieve for the size , for these "HD-like" encoding ? (of course, it depends on the duration of movie)
I think I will choose (for example for 1:45:00 time) about 1200KBit/s ( B/(P*F)= 1.82 ), so a target size of 1GB (with 128KBit/s MP3) , for a resolution of 688*384..
What do you think about that ?

Ps : I'm not totaly decided on resolution, I will make more test to see if higher res like 720*400 is better on my CRT TV screen..
but it can be nice if I'd buy one day a LCD TV ?

Brother John
17th November 2006, 18:43
I'm with celtic_druid. Just cropping and setting the PAR in the bitstream (additionally in the container, if possible) is a good solution as long as you don't care about SAP support.

My target size is usually 1/2 DVD5. For really demanding movies and/or longer than ~2.5h I started using single pass quant 2. Blank DVDs are cheap enough these days.

kurt
17th November 2006, 19:40
Some Homeplayers support AR signaling. So I'm with you guys: I always encode anamorphic.

seewen
29th November 2006, 11:26
so a target size of 1GB (with 128KBit/s MP3) , for a resolution of 688*384..
What do you think about that ?



I've nothing to say about the resolution (as it depends a lot on the DVD-player/TV too. For ex my new DVD player "toshiba" is 10x times better than my philips/kiss/eltax/Sony and produce miracles with divx, even 320x240 res.... ).

But it's not a good idea to use 128kbps MP3. It's really too low.
You should definitely use something like vbr 160-180kbps (lame quality 2 newvbr or preset-std for example) if the source is AC3 5.1, or try to keep original AC3 2.0 if it exists.

The difference between 128 and 160/180 is huge on DPL2 capable home theater. Even on average/good quality TV HP.

Dams
29th November 2006, 14:42
I really think that 128kbps CBR is really sufficient for most of movies, but I encode also with AC3 when movie style is SF or thing like that.
I doesn't encode with VBR, as I think it can produce some problem of synch ..

Brother John
29th November 2006, 15:00
The difference between 128 and 160/180 is huge on DPL2 capable home theater. Even on average/good quality TV HP.
Could you elaborate a little? In the light of Sebastian Mares's 128 kbit listening test (http://www.maresweb.de/listening-tests/mf-128-1/) I find such huge differences a little hard to believe. In that test Lame scored 4.6 with 5.0 being transparent.

seewen
29th November 2006, 16:17
I really think that 128kbps CBR is really sufficient for most of movies, but I encode also with AC3 when movie style is SF or thing like that.
I doesn't encode with VBR, as I think it can produce some problem of synch ..

CBR... :scared: it's even lower than I thought first.

VBR was causing problem a few years before, with some of the very first divx players.

Even the 30-40$ players from today play vbrmp3 correctly (and if today you find a player that is not, you could bet that it has many other problem and you won't buy it ).
( See by yourself here for example http://www.divxtest.com/form/divxtest2_list.php?lang=en .


Really you should use lame command "-V 2 -b 32 -B 320" or "--preset standard".
With series like "Friends" (no music, no sfx, many silences) you'll have something like 100-140kbps, and with "Pirates of the Caribbean" you'll have 160-180.


With a little blind test, you will hear the difference between a 128kbps CBR and a 170kbps VBR. The sound is completely different (action parts, music parts, sfx etc.)

And the day you'll buy a 5.1 home theater, you'll be even happier to have a good audio stream (there are lot of message about DPL2 decoder and mp3 bitrate in the "audio encoding" part of the forum).


And finally, what are you going to loose if you use MP3vbr@180kbps or (original)AC3@192kbs, instead of MP3cbr@128kbps, on a 2 hours long movie ?
40-45Mo !

You won't see those 40Mo missing from your video. But you'll hear a huge difference ;)


P.S.
Take one of your DVD, make a little bivx with MP3cbr@128 and MP3vbr (-V2 or --preset std).
And try to swap stream during different parts of the movie. I'm sure you'll hear differences.

seewen
29th November 2006, 16:23
Could you elaborate a little? In the light of Sebastian Mares's 128 kbit listening test (http://www.maresweb.de/listening-tests/mf-128-1/) I find such huge differences a little hard to believe. In that test Lame scored 4.6 with 5.0 being transparent.

I didn't read the test entirely.
But as you see there is no 128 CBR MP3 on this test. Average bitrate = VBR 145kbps ("bitrate used" http://www.maresweb.de/listening-tests/mf-128-1/results.htm)

And the purpose of the test is to compare differents so called "128 kbps samples". It's not to test the difference between MP3cbr@128 and MP3vbr@180, for example.

But I suggest you to use the methodology and the tools of "Sebastian Mares's 128 kbit listening test", to test the difference between original AC3 5.1, MP3cbr@128, and MP3vbr X.

celtic_druid
29th November 2006, 16:24
-V 4 or -V 5 should be fine. Much, much better than 128k CBR anyway and possibly at a lower average bitrate.

seewen
29th November 2006, 16:38
-V 4 or -V 5 should be fine. Much, much better than 128k CBR anyway and possibly at a lower average bitrate.

That's right.
-----


But again, difference between 2h mp3@128 and mp3@180 is only 40Mo. On a good quality encode, for me, you don't see the difference on video.

Dams
29th November 2006, 16:42
Ok, ok I'm from "old school", so I have to change my technique..

seewen
29th November 2006, 16:43
Tu verra tu sera bien content ^^

Dams
29th November 2006, 16:49
With a little blind test, you will hear the difference between a 128kbps CBR and a 170kbps VBR. The sound is completely different (action parts, music parts, sfx etc.)I entirely agree with you

And the day you'll buy a 5.1 home theater, you'll be even happier to have a good audio stream (there are lot of message about DPL2 decoder and mp3 bitrate in the "audio encoding" part of the forum).I've got already a good one and expensive, but not with DPL2 (5 years old Yamaha DTS/AC3 + 5.0 enclosure Elipson)


And finally, what are you going to loose if you use MP3vbr@180kbps or (original)AC3@192kbs, instead of MP3cbr@128kbps, on a 2 hours long movie ?
40-45Mo ! I agree with you too, as today, I change my way of encoding, as I decided to reach about 1000-1100MB, instead of 700 MB


You speak nice french ;)

movmasty
30th November 2006, 17:49
For myself, I prefer 688*384 which is the best "quality/price" : good resolution and not to big for frame size
uhm, lets say 704xxx, full horizontal res after cropping the black bars,
certainly you can crop a little more, 688 or 672,
cropping, the vert res will be always 400.

Sometime i like to resize back to 720, thus increasing vert res, or better, reducing it less.

Size is around 500mb for a 100min movie............12.5 fps of course :D

movmasty
30th November 2006, 17:58
I'd encode with non square pixels. That would make it around 720x544.

about this, i think that is erroneus thinking that enlaging a vid(oly orizzontally in this case),
doesnt increase quality and just waste space, still i think that better quality on playing is at zoomx2, and this mean to magnify each res in the same way.

when i was on a crt 1152x864, i made some encoding cropping to 672 and resizing to 768(768*1,5=1152), thus keeping almost all the hor res,
now on a lcd 1440x900 i prefer 720xxx, always cropping.

movmasty
30th November 2006, 18:05
720*576@720*400 is always an underresize even if you make a large crop 704*560@720*400.

720*400 is highest possible resize (SAP compliance) in the no anamorphic way if you want the best possible quality at high/medium bitrate (less than average q4 for H263 quantisation with XviD/DivX). IMO actually average q4 720*400 done better visual result than average q3 640*352.
dont agree, between Q3 and 4 there is the turnig point where artifacts comes out,
a movie encoded at q3 is clear, i think identical like q2,
at q4 it is not.

and, having space and time, you could also keep full vert resolution enlarging the hor, quality will increase, lets say 960x544 cropping something.