Swoop n' Hook: YouTube video quality

[erikph 0]

I just watched the Swoop n' Hook clip (by DSE) on YouTube.

It 's a great video, but what also struck me was the compression quality. There is some challenging footage in there, and yet it is among the best I have seen on YouTube. Few blocking or blurry spots, good colors and contrast (It is certainly better than what I can manage )

Can anybody (DSE?) elaborate on what are good settings for uploading clips, or share some tips?

thx,

blue skies,

http://myjumps.blogspot.com/

[DSE 3]

That particular encode was done with Sony Vegas, but Canopus ProCoder can get you about the same result.
Glad you feel the quality is decent.
Some tips:
~Crush blacks, reduce/clip whites. Be sure you've got levels locked in at 0-233, and use a black stretch or threshold filter to make black...black.
~Push the contrast. Encoded media loves contrast.
~Avoid lengthy fades/transitions. Encoding likes clean pixels
~Reduce saturation by about 10%. The human eye can't see that small a reduction, but the encoder sure can.
~render your master for YouTube using a 320 x 240, MP4 or 5MB WMV. Do not attempt to use MPEG, MOV of anything but MP4.
~If you're shooting DV, crop 6 pixels off both sides to clean out the DV fringing.
~consider adding .001Box blur or Gaussian blur to the master 720 x 240 before render to 320x240.

If you wanna read some old, not quite current stuff that I wrote a few years back about stream

[Squeak 17]

who shot the footage, it's some awesome flying there

You are not now, nor will you ever be, good enough to not die in this sport (Sparky)
My Life ROCKS!
How's yours doing?

[FeFe 0]

very VERY well done
which PP did you use on Z1s? or all color-correction was done on post?

[DSE 3]

The pilots are Shannon Pilcher, Ian Bobo, Jonathan Tagle, Jay Moledzki, and JC Colclasure of the Performance Designs Factory Team. They're flying custom Velocity canopies, with a mix of Wes Rich and Skysystems camera helmets. All are wearing Sony HDV camcorders.
Ground and aircraft mounted Z1's and V1's are all closely color matched using picture profiles and a vectorscope, I created custom PP's for both cam models. PP's included use of the CineGamma and Black Stretch on the Z1's and the LUT of the V1.
V1's are all time-code matched via the iLink port.
The pilots are shooting Sony A1u's, with custom settings as well, all set to 1/60, so that conformto 24p (which is part of what you're seeing on YouTube) is easier and more accurate. High framerates are not critical, contrary to popular belief, and with CMOS cams, you can run into rolling shutter probs anyway (when at high framerates).
We also whitebalanced to red cards to cool down the shots, given that warmth in the sky would have been weird in this case.
Anyway...pilots didn't have polarizers on their lenses (that's a semi-mistake, in hindsight) but any cam shooting across snow or into sky had a 4x4 polarize on it.
Other than one shot (where the image looks like molten lava, sensor FUBAR'd on the cam) there isn't any color correction. That short piece you see is pretty weak, IMO. It's a 20 minute multicam cut done so the team could quickly see what we captured. The real edits/assembly is yet to come. when it's actually corrected, balanced, etc...it should be much more visually compelling. That quick edit is exciting, but it fails to communicate the adrenaline, anticipation, and emotion that took place in that stunt. FWIW, that's one of the more weak stunts that took place. We've got some that just knocked us on our tails.

Aerials were shot by all the PD Factory Team members, JC Colclasure of course, is the Factory Team photographer, so he captured the bulk of the usable team footage.

Back to the Youtube encode, I failed to mention that the fewer frames fed to the encoder, the better the image. That's a 24fps sequence.
At the end of the day, the more redundancy, the fewer frames, less color, the better the encode.
HTH.

[FeFe 0]

cld you elaborate on TC matching via iLink? did not know it could be done... or is it V1's feature?

[DSE 3]

Quote

cld you elaborate on TC matching via iLink? did not know it could be done... or is it V1's feature?

In the HVR V1, you can set one camera as a master, connect as many V1's to the first V1 as you'd like via iLink/1394, and the connected V1's will take on the timecode flags from the master cam. Brilliant means of working with non-genlocked cams for post sync.

[FeFe 0]

I think it's to for me to buy your V1 training DVD