Downconvert question

[Peter Manojlovic]

Hey all...

I just got a last minute call for a wedding tomorrow...
The people involved don't have a clue as to formats, mixing footage, or output...

As far as i know, somebody else will be editing the footage. BTW, footage is to be put on DVD...
My heart tells me to shoot DV 16:9, but i'm sure they want everything shot in "High def".

Here's my question:


Since i'm responsible for raw DVD delivery, i'm thinking of outputting straight to a DVD recorder. That being said, how is the quality of the internal downconverting mechanism of the XHA1?
Can it be done "on the fly" with reasonable quality, or should i be looking at converting in post instead?

[Donald McPherson]

I would all ways record in the highest setting my camera can do. You can down convert easily but not up convert.

[Christopher Icha]

Always shoot High Def unless instructed otherwise.

[Brian David Melnyk]

i think quality of down converting suffers greatly from poor software scaling and have read that if delivery is sd that shooting in sd may produce higher quality results... that said, i still shoot hd anyway!!!!!

it seems bizarre that there is no definitive software approach to transcoding for dvd for best quality (at least that i have found that is clearly the 'best' quality-only the least bad!). that said, after some recent research and experimentation, i have left the HD>prores>edit>compressor 2 pass VBR (with settings from Robert Lane) and instead use HD>prores>edit>scale to sd timeline>export uncompressed sd 16:9 mov>compressor 1 pass CBR (at about 7.5) mpeg2 (*my projects are usually under 30 minutes*).

maybe compressor could also be used for the scaling to uncompressed sd step.... regardless, the logic is that compressor does a horrible job of scaling and compressing at the same time, so letting it do one job at a time produces better results. the CBR keeps a consistent quality while really reducing the time needed for transcoding. in fact, i find this process much faster, and from my clearly non-technical point of view, it produces less scaling artifacts. YMMV!!!!!!!!!!

i would love to hear some experiences from smarter people than myself!

[Peter Manojlovic]

Brian....

I've researched this and am quite familiar with the software aspect of downconverting.
Nobody (almost nobody) offers a decent downconvert, because there isn't plugin folders for different deinterlacing and reweaving the fields method.
Progressive footage gets a somewhat simple scaledown. Interlaced footage needs the fields to be bob deinterlaced, resized, reselect the proper fields, and reweaved.

Avisynth does this, but the fact that none of the professional (and fully paid) software companies have gone to the sites like doom9, or used any of the freeware code is mind boggling...

My question was for the hardware end...It just isn't worth my time anymore.

[Don Palomaki]

Delivering the raw footage on a video DVD for further editing is perhaps the worst of all worlds. (Well, I suppose you could dump it to VHS as a further step down.) The editor should capture the HDV from the original tape for best results. And I am not especially fond of the internal camcorder down conversion except for quick and dirty work or previews of raw footage. On the other hand, capturing the HDV fromtape to HDV AVI files and copying the files to a data DVD should be OK.

High qualitry down conversion is not cheap. The studios do not use $75 conversion software for preparting DVD files.

In general the best results are to shoot and edit in HD and down convert as the last step using be highest quality transcoder yo can find for the format you are using. TMPGEnc VMW5 is generally well respected among the moderate priced conversion products..

Best to speak with the editor and determine what he/she is able to use.

[Dave Barnes]

Here is my workflow...
I edit on FCP and shoot on a Canon XL-H1A and XH-A1S..

Capture HDV -> prores 422
edit , effects etc all within prores 422
output final edit to prores 422

feed prores into compressor VBR 2 pass 5.0 avg 7.5 MAX bit rate.

Results are stunning and its not bad on processor time... I just let it run overnight... my shows (plays , recitals) are usually 2-3 hours final edit.

I have found, for me, the magic key is to keep EVERYTHING in prores till fed to compressor... and let IT do all the work.... scaling/downconversion etc etc...

[Brian David Melnyk]

i have filmed with the XH-A1 HDV (now sold) and have used the HDV>prores>edit>compressor 2 pass VBR (with settings optimized by the knowledgeable Robert Lane) and also the HDV>prores>edit>output uncompressed QT mov>compressor 2 pass VBR (same settings) AND HDV>edit natively>output uncompressed mov or straight to compressor with same settings. i have also used compression markers for areas with lots of motion or small details (bricks, leaves etc.) and transitions, and have applied light sharpening in compressor. I have also output 5-10 minute segments of projects and compressed individually and reassembled in DVDSP. now i am trying HD(V)>prores>edit>uncompressed mov scaled to SD or uncompressed HD mov>compressor 1 pass CBR (once for the SD, twice for the HD mov for separate scaling and mpeg2 compression). i have also messed with GOP size and structure, anti-alias and details sliders, bit rate, resize settings etc. etc. and have only achieved increasing the quantity of time to compress, not the quality of the final product.

All resulting (to slightly different degrees) in soft footage with artifacts and blocking and a strange 'pumping' in really detailed areas, aliasing and moire. knowing the original footage, i would never describe the results as 'stunning'... sometimes i am surprised by sections that look pretty darn good, but i don't think a whole project has come out 'stunning' in its entirety.

i am pretty sure that whatever the workflow or hoops i jump through, compressor will never produce the quality i want (somebody, anybody, please prove me wrong!!!). maybe i'm too picky, but with all my inve$tment in HD cameras, equipment, and workflow to achieve the highest quality i can afford- why has software not been developed/included that puts a really 'stunning' final product into the hands of professionals outside of Hollywood? is there some technical reason i am missing? (note: i live and work in Mali, so no access to hardware conversion facilities...)

[sorry to hijack the thread... but i agree with the advice to send the editor the HDV for most options for higher quality output, though maybe i don't understand why you have to deliver the footage on DVD and at SD...]

[Dave Barnes]

Hi Brian..

>>why has software not been developed/included that puts a really 'stunning' final product into the hands of professionals outside of Hollywood? is there some technical reason i am missing?
Yes I understand your point here... and over the years have wondered this myself...



Perhaps my use of 'stunning' was a bit too harsh... :) Anyways I am VERY happy with the output in SD for DVD results that I get with my outlined workflow.... One thing I can say I learned (the hard way) .. and I am sure you know this... back when I started using compressor 1.0 , when at that time I was still shooting S-VHS on my JVC GY-X2B (very nice camera for its time) , I discovered that the cleaner, sharper the starting image the better compressor could do its job... when I jumped to DV (Canon XL2) the same held true... then when I jumped to HDV , I realized quickly that focus is now CRITICAL... and some experiments early on resulted in some pretty bad , or blah looking SD dvd's. At the end of the day the more pristine the SOURCE you feed compressor the better it can do its job... it may be slow... but it CAN produce some really VERY GOOD results...

So to the original post... yes I agree get the original untouched HDV footage into thier hands for editing...

[John Estcourt]

Hi Peter, I hope you don't mind me asking, but how did you get on and what format did you shoot in?