Welcome to our community

Be a part of something great, join today!

Register Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Best Quality Hardware or software downconverter?

G

Gabriele Turchi

Well-known member
Joined
Mar 10, 2008
Messages
844
Reaction score
0
Points
0
Location
New York City
I want to produce the best H264 SD or DVD SD as possible out from my HD masters in SCRATCH ,so the first step for sure is a good Scaling downconversion ,
so
Speaking quality of scaling (nor artifacts and as sharper as possible)

which one of this products can be consider an high quality downconverter like the scaling quality of decks like D5 or SR?

Teranex Mini
Aja HD10MD3
Aja Kona 3
Blackmagic upcrossdown
Bluefish Epoc (whit jet stream module for downconversion)

or

Any Top end quality software solution?

Thanks a lot

g
 
Try This, Id be interested to see how it compares with other processes youve tried.

Use Apple Compressor. 3 or later.
Enable Frame controls and set up suitable parameters for down convert. ie size etc. (Theres not a huge difference in the 'better 'and 'best' options for spacial down convert.) This component is the same high quality component used in HSake which is very good.

Download the X264 QT encoder component and install. (just google for it and you'll find it. Set up suitable parameters in the X264 encoder, this isnt easy but give it a go, we may have some presents we could mail you if needed.

Hope it works out.

Dave
 
thanks dave ,

but i've used compressor with frame control many many times , so far i can say that is better than a Kona LHe (and Kona 3 and blackmagic chep converter ...)

i think that the teranex and decks (and even teh bluefish )should do a better job...the problems on this cheap solutions is the aliasing on the edges ...

i am sure somewhere must be a software that does a great job , great as decks...

g
 
If speed is your concern then you need hardware, but often software can be higher quality, although there are three basic types of software downsampling

a) software that aims for pure speed at the expense of quality
a) software that makes a compromise between quality and speed
b) and software that can focus purely on quality output

Ask yourself this: Is output quality better with or without a rocket?

If you think DRX can improve your image, then the answer is "without", as its too complex for the redrocket hardware... the beauty of non-realtime software is, it can be as complex as you like.

I wrote my own downsampler for image sequences ( and recently added the ability to encode to DCI jpeg2000 files as output). I don't think anyone really "got it". It can downsample anything, regardless of footage subject or resolution. Its all floating point beauty

eg, 28k epic stock footage of herring bones caught in a spiders web -> CIF (176X144 pixels) lol! .. no problem.

I'm personally not obsessed wth speed, you may not have the luxury. That said, the bottle neck with my system now is almost purely disk I/O speed/space, not the algorithm itself.
 
Thanks conrad ,
speed it's not an issue for me , i want quality when i have to deliver a job As DVD ...

Is Your software on the market?
Any chance to' try it ?

Ps:johhny ,i di have the matrox max encoder ,and i agree, for h364 is really great...

G
 
Gabriele Turchi said:
Thanks conrad ,
speed it's not an issue for me , i want quality when i have to deliver a job As DVD ...

Is Your software on the market?
Any chance to' try it ?

G
Click to expand...

I believe I have an already compiled version suitable for dvd output. Its Windows software though. It runs on anything..and I mean anything (ie, windows 3.1 with win32s installed.. which hopefully means nothing to you)

It loads image sequences, downsamples them at full quality (floating point), then outputs as 8bit bmps or pngs (perfect for dvd, which is what I used it for). It doesn't suffer from out of phase patterns on shirts etc

My newer version would need digging/sorting out a bit (GUI-wise, it needs making more user friendly). It differs from the above version in that it can load and export more formats by calling/using third party commandline tools, and can output 16bit images. I'm not editing at the moment, just renting gear and playing guitar so development has/had been paused.

pm me your email if you want the dvd version

UPDATE: Heres a basic GUI for some tests I've been doing. Output goes into same folder as source images, as do temporary files. I've only enabed TIFF input and output at the moment, and they are read and written as 48bit files. Its the result of a busy night this week. It can read and write most formats either internally, or using external tools, when I enable it.. when I get time.

There is a large test chart (uncompressed TIFF) in the program folder. Move it to another folder and process that folder using the default output sizes (273x341 I think, 1/100 the size of the aforementioned TIFF) and process. No go try the same resample in PhotoshopCS4/ AfterEffects etc. I did ;)

here's the link:

www.northyorkshirecamerahire.com/tds48bit.zip
 
I went through this quite a while ago and tried every software and hardware solution I could find; in the end I discovered a rather unlikely solution, the Panasonic DVCPro HD deck. No kidding; this produced the best looking down-conversions. If you have an old one lying around, give it a try; just run the signal through the electronics.

I think Kona and Black Magic do a good job these days.

Good shooting and best regards,

Leo
 
Leo Ticheli said:
I went through this quite a while ago and tried every software and hardware solution I could find; in the end I discovered a rather unlikely solution, the Panasonic DVCPro HD deck. No kidding; this produced the best looking down-conversions. If you have an old one lying around, give it a try; just run the signal through the electronics.

I think Kona and Black Magic do a good job these days.

Good shooting and best regards,

Leo
Click to expand...

i offered you mine but you weren't interested! The guy wants quality. not speed. Hardware doesn't come close. I can't wait for 9k/28k+ footage to arrive :lol:
 
Heres a basic GUI for some tests I've been doing into downsampling. It downsizes TIFF images. Output goes into same folder as source images, as do temporary files. I've only enabed TIFF input and output at the moment, and they are read and written as 48bit files. Its the result of a busy night this week. It can read and write most formats either internally, or using external tools, when I enable it.. when I get time. That code is already in there, including encoding to AVI/ MPEG etc, I turned it off until i can test everything. Eventually I intend to allow keyframed imagepans across a sequence of TIFs using splines with continuity (so looping imagepans can be created), and HQ upsampling/super-resolution of sequences... and very soon the ability to choose a different output folder!

There is a large test chart (uncompressed TIFF) in the program folder. If you want to test the downsampling, move it to another folder and process that folder using the default output sizes (273x341 I think, 1/100 the size of the aforementioned TIFF..) and process. No go try the same resample in PhotoshopCS4/ AfterEffects etc and compare results. I did ;) There is a complex mode which yields insignificantly better results..so don't use it.. its just a test mode really, and very slow. best regards

here's the link:

www.northyorkshirecamerahire.com/tds48bit.zip
 
I've done a lot of research into high quality down-conversion for our broadcast output and would suggest using a HDCAM SR deck for this or the Snell and Wilcox Alchemist.

The S & W guys are at the top of their game and I believe they are looking at a software solution but I'm not sure when this would be released. I spoke with their R & D guys and they explained that software down-conversion should always be better than hardware as this does not have to work in realtime.

The BBC here in the UK recommend the SR deck option and they are VERY fussy.

After Effects does a good job and Compressor is also decent but slow.

Hope this helps.

Best wishes

Scott
 
Scott Brown said:
I've done a lot of research into high quality down-conversion for our broadcast output and would suggest using a HDCAM SR deck for this or the Snell and Wilcox Alchemist.

The S & W guys are at the top of their game and I believe they are looking at a software solution but I'm not sure when this would be released. I spoke with their R & D guys and they explained that software down-conversion should always be better than hardware as this does not have to work in realtime.

The BBC here in the UK recommend the SR deck option and they are VERY fussy.

After Effects does a good job and Compressor is also decent but slow.

Hope this helps.

Best wishes

Scott
Click to expand...

They're right! And you're right about After Effects.. especially compared to Photoshop.. but you should try my downsampler, then send a link to the S & W guys and then the BBC ;) ..there's a test chart in the .zip for testing and comparing..

aftereffectscomparison.png


UPDATE: I've re-written it for a fourth time. This one seems to have a stable GUI and can export in different formats (TIF,DPX or J2K, all 48bit). You can also choose a different destination folder, and despite using less memory overall, it now uses double precision floats instead of single precision. same link as above
 
Last edited:
Would you kindly repost the link ? the link above is down .
 
I personally use AVISynth's spline36 scaler for downsamples to HD resoltuions, and Lancos4 for Downsamples to SD resolutions. This can be very slow mostly because disk IO isn't as fast a real time on a tape deck.

That being said, most tape decks have very good hardware downsampling. In my experience the BlackMagic Multibridge Pro is a good SD downsampler, but that doesn't get you to mpeg-2, so there is still another step to do it that way.

AviSynth feeds straight into CCP and HCEnc, 2 of the best software Mpeg-2 encoders available. Depending on your comptuer setup, this could be a faster solution.

By the way, scaling methods are a subjective matter. I don't think it's fair to say there is a "best" method.
 
Most of the good methods look almost the same if you ask for downscaling. Downscaling progressive is very easy, downscaling interlaced is yet another challenge- it's not as easy as to make it as sharp as possible :)
Upscaling is a different matter and nothing comes close to avisynth- including Teranex etc.
 
You must log in or register to reply here.
Back
Top