What percentage of first pass bitrate should be used for second pass?

[mgh]

If I enable encoding status under advanced options and do not select autoclose once the tab is open during encoding, I get a calculated bitrate at the end of the first pass. (default settings using either HT PAL profile or HT NTSC profile).
Question-what is the guideline for selecting second pass bitrate for playback on a twentynine inch CRT TV?

What i am asking for is a thumbrule-similar to- 90 mins max per CD,longer movies go to 2 CDs.
Is it possible to have such a thumbrule?

[Heini011]

hi,

>what is the guideline for selecting second pass bitrate for playback on a
>twentynine inch CRT TV?

damn! i know only the percentage of first pass bitrate/filesize for 13th century adventure movies with 14% action and 3% love scenes played on twentyone PC TFT. ;-)

to be serious: it depends for example on the desired quality and the amount of motion and noise. a movie with much motion and/or low quality can be compressed stronger.

my impressions for an average, well denoised/noiseless movie...

second pass size against first pass size:

- (0..10) % barely visible quality loss
- (10..20) % slightly quality loss
- (20..30) % appreciable quality loss
- (> 30) % apparent quality loss with blocking on motion

greetings.

[yaz]

my impressions for an average, well denoised/noiseless movie...
well. but most of us live in a world where such sources do not exist, so i'd recommend to multiply the limits by 2-2.5 giving sg like : 0.33 - 0.5 - 0.66. in general, i wouldn't recommend to go below 50% compression w/real sources.
of course, it's just a (very rough) rule'o'thumb. in some instances, the compression can be much lower, but most of the clips i've ever tested gave me unacceptable quality by applying more than 50% compression.

the bests
y

[mgh]

At the end of first pass, I get bitrates from about 1500 kbps to 4500 kbps for a 640x480 encode, depending on complexity of the video. Bitrate for .2 bits/ pixel (normal recommended value) gives 1500 kbps for 25 fps.
My thinking was that the first pass value could be used to get an idea of the final bitrate needed. The reason for mentioning the TV size was that many people appear to be of the view that above 29" requires muvh higher bitrates.

[Teegedeck]

Such rules of thumb are quite unreliable - sources tend to react differently to compression not only in terms of size but also of in terms of 'look'. And there is always the factor of individual 'taste'. Personally, I would not want go for anything worse than an encoding that averages around, like, quant=3.6 (using SixOfNine) at full anamorphic resolution - but ideally it should be 3.0. You on the other hand might find higher average quants or different matrices or lower resolutions absolutely acceptable. Test it out and encode accordingly. Like:

Use the tool called 'Enc' to make a filesize prediction at the constant quantizer and picture resolution that you regard as 'good enough' for your taste.

If the predicted filesize is smaller than what you aimed for increase the resolution. If you already encode at full resolution choose a custom quantization matrix designed for higher bitrates (like SixOfNine).

So then. Of course this puts it very simply, but you might understand what I mean.

[Heini011]

Hi.

@ yaz,

>but most of us live in a world where such sources do not exist

i don't know where you live, but i live in a world, where high quality dvd's and filters such as mvdenoise (i use it always, even on good sources) exist. the point is: is someone like it cheap and fast or is looking for a high-quality restoring...

i don't understand why somebody encoded today at vga-resolution only, where dvd+/-r's are cheap and the needing for a higher resolution (hdtv) in general is apparent. but if doing so, quality loss caused by to low bitrate is particularly eye-catching.

[Didée]

i live in a world, where high quality dvd's and filters such as mvdenoise (i use it always, even on good sources) exist.
Carefully - look out for objects illuminated by flaring fire, light-through-ventilators, and such.

[MeteorRain]

IMHO, usually 2nd/1st = about 70%