Final Cut Pro Mojo

[John Sagrelius]

[Oliver Peters] “The only thing Xpro/Mojo enables that FCP/Io doesn’t is mixed codecs in RT (uncompressed + DV) on the same timeline.”

You’ll find some interesting feedback regarding FCP and mixed codecs in this thread: https://www.creativecow.net/forum/read_post.php?postid=111652268014838&forumid=124

[Oliver Peters]

[MPE] “As you have said, the codec is on board. Thus some of the intensive tasks of compression/decompression is off loaded….
…..Try using Io for DV. Decent amount of RT as long as you keep it in Unlimited/Low Quality. ….
….With Adrenaline – full quality 4-8 layers of DV. Mojo the same but of course limited to 8 bit.
..I understand it Mojo does capture DV native via its firewire port if you set it to DV. …”

MPE,

You are mistaking optimization of the software for the CPU/GPU with what the external system does. Avid works the same way in performance with or without either Mojo or Adrenaline attached. The difference is that Adrenaline acts as a dongle for MC so you can’t even make an assessment without Adrenaline connected. Mojo acts as a dongle for 1:1 on Xpro. You can see this if you run XproHD without Mojo and work with DVCProHD and DNxHD. Works just fine. No Mojo required.

In the case of Io there does seem to be some overhead lost to feeding the output, so I can see where performance might actually drop a bit. Aja is touting acceleration when you use Kona2. This is an installed PCI card which is quite a bit different than Io, Mojo or Adrenaline.

The onboard codecs of Mojo do nothing other than handle I/O tasks. If there was any actual acceleration going on, the computer would be having to send MULTIPLE streams of data out through the FW port to the Mojo and then back again and then one more time for a composited output. That simply isn’t happening. You’re correct about the native DV on Mojo, but this also means you have to transcode anything that isn’t DV down to DV. This means all your 1:1 media gets compressed to DV. This is RENDER process which does not gains any RT performance from Mojo.

Sincerely,
Oliver

Oliver Peters
Post-Production & Interactive Media
Orlando, FL
http://www.oliverpeters.com

[Dom Silverio]

[Oliver Peters] “Avid works the same way in performance with or without either Mojo or Adrenaline attached.”

[Oliver Peters] “The onboard codecs of Mojo do nothing other than handle I/O tasks. If there was any actual acceleration going on, the computer would be having to send MULTIPLE streams of data out through the FW port to the Mojo and then back again and then one more time for a composited output.”

I think you are incorrect. You are looking at this at a traditional way a PCI accelerator board architecture would – send multiple streams and let it be sorted it out. DNA does it slightly different [except for Nitris] or at least its duty is not so early in the process. All the computative tasks is handled by the CPU. However, the recompression is done by the on-board codec. So for example, for DV with 3 PiPs, the computer will calculate the FX (in uncompress mode is the popular theory since it has the lowest overhead while maintaining the best quality) and send an uncompress stream down to your DNA. The DNA in turn recompresses it back to DV format. The result? More stream of DV since the recompress is not handled by the CPU.

You can test this theory with XPro version 5. It has real time DV out via firewire in full quality without Mojo. With a 3 GHz P4 and 1 Gig of RAM, I can push about 2-3 layers. Pop the Mojo in and 4-5. With faster CPU [say P4 3.4 GHz and more RAM] I can push more streams – 3-5 without Mojo and 5-7 with Mojo.

So does that mean there is no acceleration with uncompress? Probably not. This is why FCP and Adrenaline have relatively the same amount of 1:1 8 bit stream with Adrenaline slightly ahead on the HP [probably caused by CPU difference and not NLE architecture difference]. But for compressed formats like DV, 2:1, DVC Pro HD, DV50, etc – the DNA does perform more streams- at least enough that it is difficult to explain by platform variation.

.02

[Oliver Peters]

[MPE] “All the computative tasks is handled by the CPU. However, the recompression is done by the on-board codec. So for example, for DV with 3 PiPs, the computer will calculate the FX (in uncompress mode is the popular theory since it has the lowest overhead while maintaining the best quality) and send an uncompress stream down to your DNA. The DNA in turn recompresses it back to DV format. The result? More stream of DV since the recompress is not handled by the CPU. …
…..But for compressed formats like DV, 2:1, DVC Pro HD, DV50, etc – the DNA does perform more streams- at least enough that it is difficult to explain by platform variation.”

MPE,

I think we are actually in agreement as to how the system works – we just differ on the symantics. What you describe above is exactly what AJA is doing in Io, with the main exception that Io does not recompress to DV or anything else. Output is only uncompressed (analog or digital). Io has no pass-through DV ports, so if you want to work in native DV25/50 from the deck, you have to disconnect Io. IoLA is more or less equivalent to Mojo and the full Io is more or less equivalent to Adrenaline. The question is whether Adrenaline is justifiably worth an 8X larger price tag.

Sincerely,
Oliver

Oliver Peters
Post-Production & Interactive Media
Orlando, FL
http://www.oliverpeters.com