No this is not just a shameless plug for my 3 month old LH & KBox bought to do one 1920x1080i60 HD project, and now available on ebay here with the full balance of the three year warranty transfering to the purchaser:
http://cgi.ebay.ca/ws/eBayISAPI.dll?ViewItem&item=7586490858&rd=1&sspagenam...
:-)
I was faced with a one-off HD 1920x1080i60 for broadcast long form documentary that had footage shot on DV, DVCam, XDCam, HDV, and HDCam. It also had Betacam SP stock footage. The producers were very worried about how it would all look when put together, and had concerns with broadcaster acceptance.
Although there are some other differences, the choice of LH or Kona 2 came down to two issues:
1-The Kona 2 could do upconvert, but not analog component HD, which is the best picture you're gonna get out of HDV originals.
2-The Kona LH could not do upconvert but could do analog component HD avoiding the "issue" of HDV altogether.
The decision to go LH was made because:
1- all advice said that the Kona 2 upconvert gave varying results, sometimes great and sometimes perhaps not acceptable. It was suggested by all my research that way to go for a no compromise upconvert was the Teranex. Teranex was what we ultimately used for all of the SD material
2-the 12 bit analog component HD out of the camera kicked the crap out of the HDV out.
Rendering an upconvert in FCP did not have good results. Oddly, also rendering a downconvert in FCP did not have good results. The hardware downconvert in the LH was great.
The Online...
The program was treated as two separate onlines. The first was done on all of the SD material, leaving blacks for the HD material and handles for effects on the SD material. This was colour (Canadian spelling eh!) corrected and conformed in SD 10 bit uncompressed, put to Digital Betacam via SDI, and sent for Teranex to be upconverted to HDCam.
The Teranex worked great. The settings are done shot by shot with the settings based on the technicians interpretation of the footage. The upconvert is then done in real time after all of the settings are done. Allow 1.5x against your total showtime for your booking. For now, Teranex seems to be the undisputed champ of upconvert.
Once this was done the HD part of the job moved forward. The HDCam material was digitized via SDI. The HDV material was digitized via component HD, and the Teranex material was recorded in from HDCam via SDI with the timecodes of the Teranex(ed) material already matching the master program timecodes, No muss no fuss.
It was surprising how well the HDV original performed when digitized analog component. Using this method the footage was indistinguishable from the 8 bit HDCam. Most of the material in this show was fairly static, so the HDV recording codec was not under much load motion-wise - so it wasn't a definitive test. Anyways, the results I saw certainly wouldn't have me jumping out to get an HDCam for $50k more over HDV for 1080i60 work.
Drives, 8 or 10 bit.
I did the colour correct, and prefer to work in 10 bit. I feel the difference is huge, especially if you have to blow anything up. Everything holds up better and the shading is much richer. Even though the delivery was 8 bit HDCam, I digitized all in 10 bit uncompressed 1080i60 for the colour correction and HD conform. If your show is going to be a film print one day, then I would definitely go 10 bit.
I was able to digitize and work fully in 10 bit uncompressed on an 8x400 drive SATA RAID with a dual 2.7 and 6Gb RAM. I used two Sonnet SATA controllers. The slot configuration was: LH nearest the video card, with the two SATA controllers in the next two slots. The slot farthest from the video card is the fastest one. The 8x400 drive RAID had 4 drives striped RAID0 across the two controllers. I had a second 4x400 RAID set up for bouncing renders. 12x400 in all It all worked great.
I was not, however, able to output 1080i60 10-bit Uncompressed on this RAID. It was not quite fast enough, and adding drives to the RAID did not help.
The RAID did handle an 8-bit 1080i60 uncompressed output which is all that was required for HDCam work anyway and everybody said "spectacular" when they saw the finished show.
If the output had been to HDCam SR, which is 10 bit, I would have had to have a faster RAID . I also could have sent the entire finished program (around 600 Gigs for 10 bit 1080i as a Quicktime movie on a hard drive for output.
My experience with this would make me hesitant to believe a 5 drive SATA RAID would work well for uncompressed HD (1920x1080i60). I don't think any 7200rpm five drive raid could do it, no matter how much bandwidth on the host card(s).
If you don't want any nasty suprises at the end, the best way to practically test a RAID is to fill up your drives with media, make a big sequence, turn of mirror video on desktop, and do a test print to tape at full quality. If FCP won't do it the RAID isn't fast enough.
In the end, this show worked out great, and everybody was very happy. It pretty much went as planned.
I think there's some nuggets in there for you...and good luck with it!
PS. the only reason I'm selling the LH is because I have an io and won't be seeing another HD project until perhaps late in the year...the LH worked great!
There's also the RAID and a Final Touch SD available listed with the same seller.
http://search.ebay.ca/_W0QQsassZliz8954QQhtZ-1
Reply
Like
