Rendering times

[Fred Foronda]

In rednering DV I get about 1:2 ratio in rendering. In other words for every minute of edited footage it would take two minutes to render. (4 min clip would take 8 min). What kind of figure would I get when it comes to rendering HDV?? Or what kind of times are you guys getting and please list the spec of your computer so I can get an idea how it will be handled on my comp.

thanks

[Steve Crisdale]

Originally Posted by Fred Foronda

In rednering DV I get about 1:2 ratio in rendering. In other words for every minute of edited footage it would take two minutes to render. (4 min clip would take 8 min). What kind of figure would I get when it comes to rendering HDV?? Or what kind of times are you guys getting and please list the spec of your computer so I can get an idea how it will be handled on my comp.

thanks

Rendering to what format? I've found there's a few variables that can make a difference to render times, and generally render times go up dramatically with the amount of processes being utilised.

The workflow used can also play a part in render times. Use of proxy versus uncompressed... So many variables!!

I've got a P-IV 3.2Ghz HT 800Mhz FSB, 1Gig RAM, 3x120gig 7,200rpm HDD (2 HDD in RAID-0 array), and I've experienced renders with ratios more like 1:10 with some HDV to WMV9 HD projects. Some renders to m2t (for writing back to my cam) from CFHD have been a lot less painful... more in the region of 1:4. HD involves a lot more infromation than DV, so you must expect a performance penalty in dealing with it - just don't expect anything approaching 1:1 from any machine that isn't a Dual-Core!!

[Kevin Shaw]

Here's what I'm getting on a 3.0 GHz dual-core system using Edius Pro 3.5 with Procoder Express:

HDV in Canopus HQ format to widescreen SD MPEG2: ~1.5-2 minutes per minute of timeline.

HDV in Canopus HQ format back to HDV 1080i: ~6-8 minutes per minute of timeline.

HDV in Canopus HQ format to Windows Media 720p: ~10-12 minutes per minute of timeline.

HDV in Canopus HQ format to Windows Media 1080p: way too long, but it looks great.

Bottom line is that you need patience to deal with HDV output renders, and perhaps even a separate computer just for that task.

[Douglas Spotted Eagle]

Originally Posted by Fred Foronda

In rednering DV I get about 1:2 ratio in rendering. In other words for every minute of edited footage it would take two minutes to render. (4 min clip would take 8 min). What kind of figure would I get when it comes to rendering HDV?? Or what kind of times are you guys getting and please list the spec of your computer so I can get an idea how it will be handled on my comp.

thanks

Fred, it's not possible to simply take two identical hardware systems and expect the same results, aside from the render times others may post. Render times will vary depending on what's running in the background, what hardware is prioritized, how the system is tweaked, and umpteen dozen other factors.
While you can get roughly in the ballpark, that's about all you can do. Look at what you're seeing in the below posts from two different apps; 1:4, and 1:12 on the outside. What filters applied, what compositing, how tracks are managed...it's all relevant.

[Kevin Shaw]

Originally Posted by Douglas Spotted Eagle

Look at what you're seeing in the below posts from two different apps; 1:4, and 1:12 on the outside. What filters applied, what compositing, how tracks are managed...it's all relevant.

If you read the posts closely I think you'll see the difference isn't quite that great: for rendering to M2T we're seeing ratios from 1:4 to 1:8, and for rendering to Windows Media 720p it's ratios from 1:10 to 1:12. So we can at least see some approximate trends, and we can definitely conclude that rendering HD output is going to be time-consuming until we get faster encoding solutions.

An interesting question we can ponder soon is, if it takes a lot longer to encode to Windows Media or H.264 than it does to encode to M2T, and we can store almost two hours of M2T on a blue-laser disc, why bother to use the more heavily compressed formats for HD video delivery?

[Fred Foronda]

Originally Posted by Douglas Spotted Eagle

Fred, it's not possible to simply take two identical hardware systems and expect the same results, aside from the render times others may post. Render times will vary depending on what's running in the background, what hardware is prioritized, how the system is tweaked, and umpteen dozen other factors.
While you can get roughly in the ballpark, that's about all you can do. Look at what you're seeing in the below posts from two different apps; 1:4, and 1:12 on the outside. What filters applied, what compositing, how tracks are managed...it's all relevant.

I understand all those different factors I just wanted to see what everyone is getting out of HDV. Its a huge chunk to render, I too want improvements as far as making it faster not by just getting faster computers but as Kevin stated faster encoding solutions. Thanks for your input guys!!

[Douglas Spotted Eagle]

you do understand that the lesser the compression, particularly uncompressed, are fasster to render than any of the compressed formats, right?

[Fred Foronda]

Originally Posted by Douglas Spotted Eagle

you do understand that the lesser the compression, particularly uncompressed, are fasster to render than any of the compressed formats, right?

Using less "outsiders" to encode HDV on vegas example: Cineform, will tad the process up faster? Or as I understand it the less you mess with the original file the better the render time. Forgive me, if HDV101 was a course I'm failing right now.

[Steve Crisdale]

Originally Posted by Fred Foronda

Using less "outsiders" to encode HDV on vegas example: Cineform, will tad the process up faster? Or as I understand it the less you mess with the original file the better the render time. Forgive me, if HDV101 was a course I'm failing right now.

You're getting the drift!! No failing there at all.

Remember that re-rendering/downconversion etc. require the encode algorithm to 'determine' where a given pixel must be rewritten having read a number of given pixel data from the source file.

Just imagine having a 4x4 cell box -16 squares of different colours, and you have to now make the 16 fit into a 3x3 box... how you going to do that while retaining as much of the possible original information?

Think about it - there's lots of possible permutations... that's what we're asking the software to do. Of course the faster the hardware is the faster the software can do it's thing.

Generally, you can reduce render times by reducing the complexity of what you're asking the encoding algorithm to do.

If all the encoder needs to do is take a look at the data, recognise there's not much to rebuild - like in a 1:1 type function, then things should work a bit quicker.

We assume of course that you know how to turn off all OS background functions, virus scanners, back-up utilities etc., etc. as these will hamper performance significantly also.

As Douglas said; there's more to the picture than just "Oh, I'm seeing 1:8 render times"... while I can concur with Kevin in that there's certainly a performance hit for certain types of renders/encodes, which it's worthwhile knowing about - specifically the WMV9 one!!

Is there really a HDV101 course? God... is HDV spreading quick or what!!