Of course I would rather have 4:2:2 10-bit, but at this point it seems like that's far away so I would be happy recording the minimum of 4:2:2 8-bit, which is still better than 4:2:0. I would like to use the 7Q as something other than just a monitor.
Odyssey7 & Odyssey7Q Status Update
- Thread starter Mitch Gross
- Start date
Of course I would rather have 4:2:2 10-bit, but at this point it seems like that's far away so I would be happy recording the minimum of 4:2:2 8-bit, which is still better than 4:2:0. I would like to use the 7Q as something other than just a monitor.
hugofilipe
Member
Hi there,
I've lots of hope on you guys (CD) for future development, so I'm looking to buy the Odyssey 7Q for my FS700.
Any recomendations of where to buy in Europe?
Thanks
I've lots of hope on you guys (CD) for future development, so I'm looking to buy the Odyssey 7Q for my FS700.
Any recomendations of where to buy in Europe?
Thanks
This is just conjecture, but I don't think it has anything to do with misplaced priorities. You can already get uncompressed 8-bit 4:2:2 to ProRes out of the camera through HDMI with a third party external recorder. Most tests I've done comparing native AVCHD 8-bit 4:2:0 to HyperDeck 2 footage show such minimal improvement, that I think the extra weight + SDD and hassle of powering/mounting the HyperDeck are not worth the effort.
At this point, since the 2K RAW is pixel-binned by Sony before it gets to the feed, the only major benefit the O7Q can offer is some de-bayered/down-rezzed version of the 4K RAW stream. And the sequence of programming steps I believe are necessary to use the 4K stream:
• Decode/Debayer 4K Sony RAW (with Sony's specs and mix of guesswork)
• (optional) encode debayered 4K 4:4:4 data into 10-bit 4:2:2,
• (optional) downscale 4K into 1080p
• Write final format to compressed codec
• Certify any specifications that need certification
• Do this all in real-time, with fault tolerance
From a programming perspective, sounds like a pretty large task to me. Given CD's background with the nanoflash and Gemini, I'm pretty sure their engineers know a thing or two about video compression. But maybe this is a new set of CPU/power/bandwidth/memory mix with all sorts of fun nuances. Who knows.
My guess is that the bottleneck is dealing with the 4K RAW stream. Perhaps they're fine-tuning the debayer and color science. Regardless, the 4K hold up (theoretically), would slow down all the other features from moving forward, including 1080p. And I'm pretty sure the engineers are going full steam on getting that to work, so I don't think it's a priority issue at CD. Just a time issue.
You're not missing out. You can already record 4:2:2 8-bit uncompressed out of the stock HDMI. 4:2:2 8-bit really not much better than 4:2:0 8-bit, especially when keying. Not worth the hassle from my tests. 10-bit is what we really need.
cuervo
Veteran
Exactly my point. I mean, how much effort is needed to provide the rudimentary 422, 8 bit? I don't want or need Prores, or "some popular codec". I'll take DNx220x, as something right now. I have a PIX 240, which is a very fine DNxHD and Prores recorder. The only advantage to the O7Q is a more efficient form factor and a better display. But, that's not enough to make me want to carry BOTH, just so I can have this sexy monitor. For the O7Q to be worth anything close to what I paid for it, it needs DNxHD, like they advertized. And the OP1a variety would be even acceptable. The most fundamental of all recorders, the BMD Hyperdeck has it. And to REALLY sweeten the pot, I'll take a 4K derived XAVC. THAT would make the wait worthwhile. I'm really surprised the O7Q didn't ship with something other than DPX. That was a pretty poor choice, IMHO.
Last edited:
"How much effort?" Presumably quite a lot. This is Convergent Design, they already have programmed for the nanoflash and Gemini. But I bet you the CPU and programming for the O7Q was on a brand new sexy chip and porting code can take a lot of time. Sometimes porting code is cake, other times it becomes the bain of all existence and a bigger PITA than ever expected. Let me re-frame the question for you: how much effort does it take to shoot a feature film? Ask some non-film folks and be prepared for some fun answers.
The question shouldn't be "How much effort?", but is Convergent Design doing everything they can? Based on the fact they had folks assembling during Thanksgiving and Mitch is on this forum answering questions non-stop, I'd more than say heck yea!! To be honest, they could've very well pulled a Black Magic and just NOT have released the O7Q until Feb 2014 with all the features enabled. Instead they 1.) Listened to customers 2.) Dropped prices(!!) 3.) Hired Mitch 4.) Rushed to get the product into folks hands 5.) Announced new features. To me all indicators that they're working 110% at delivering what they promised.
I believe the O7Q will eventually have all the features they promised and then some. It will just take time, welcome to the early adopter club. If you're upset with your purchase, I'm 100% sure you can sell it for full retail price right now considering they're on back-order through March. Then just buy it again when the software is ready for you.
Mitch Gross
Veteran
I'm not going to intrude in this conversation too much. You all have strong feelings and we knew the reactions would be passionate when we prepared the update. It's understandable.
Let me state that we are not working on an 8-bit solution. 8-bit is yesterday's news and we have no interest in it. You can always truncate your material later if you wish to. That's not what the Odyssey7Q is about.
Certain things take longer than expected. Some of them much longer. Our priorities were to get out a certain set of features on first release. Then as some deadlines passed by, we had to make some decisions on what we would continue to wait for. Think of all the things the RED One, Alexa, Epic, F5, F55 and on and on did not do on initial release. But no excuses -- it's taking longer than expected even with hiring additional staff. We're working as fast as we can but these are not plug 7 play functions. You don't just drop them in and that 8-bit codec is just as much work as the 10-bit one (more if you want to insure it is the "right 8-bits").
There are certain matters on which I simply cannot go into further detail. But we do believe that we will be able to deliver you all that we have promised and are working to do so as quickly as possible.
Let me state that we are not working on an 8-bit solution. 8-bit is yesterday's news and we have no interest in it. You can always truncate your material later if you wish to. That's not what the Odyssey7Q is about.
Certain things take longer than expected. Some of them much longer. Our priorities were to get out a certain set of features on first release. Then as some deadlines passed by, we had to make some decisions on what we would continue to wait for. Think of all the things the RED One, Alexa, Epic, F5, F55 and on and on did not do on initial release. But no excuses -- it's taking longer than expected even with hiring additional staff. We're working as fast as we can but these are not plug 7 play functions. You don't just drop them in and that 8-bit codec is just as much work as the 10-bit one (more if you want to insure it is the "right 8-bits").
There are certain matters on which I simply cannot go into further detail. But we do believe that we will be able to deliver you all that we have promised and are working to do so as quickly as possible.
Mitch Gross
Veteran
It is in our price list, but with the introduction of other models we decided to modify the product code. Below are the battery plates we now offer, available through any of our dealers worldwide. We just announced these new models and many of the dealers seem to have missed the first battery plate release, so they might not appear on anyone's website. But if you contact them directly they can absolutely order them from us.
CD-OD-SLplate (Sony L series battery plate)
CD-OD-SUplate (Sony U series battery plate)
CD-OD-CBPplate (Canon BP-9xx series battery plate [C300 video camera batteries])
CD-OD-PCGAplate (Panasonic CGA series battery plate)
CD-OD-JVCplate (JVC battery plate)
paulcurtis
Veteran
I'm a big proponent of what CD are doing, i cannot think of a more open and progressive company right now. I'm also aware that development is often a rocky road.
I personally (IMHO) feel in retrospect that 2K RAW could have been a bit of a diversion in terms of resources and if 4K could have been first then it would have introduced 1080p and 4K solutions upfront and the HFR 2K RAW could have been further down the list.
Taking a straw poll i doubt that many are using HFR day to day (there are some i know). But anyway for whatever real world reasons we're in this timeline now.
Just to clarify however
It sounds like there might be *no* recording from the internal HDMI/SDI output to 1080p compressed? (The comment about the 8 bit solution). Therefore the O7Q wouldn't be a generic substitute for a hyper deck shuttle, recording from an FS100 or other HDMI/SDI cameras?
For me, just being able to record the normal signal would have been a stop gap, enabling me to re-rig and swap the O7Q in for the time being, i'm quite surprised i can't.
Reading between the lines it does seem that perhaps the popular codec might be introducing delays. Having said that i've had no end of problems with DNxHD so i don't see that as an alternative. But pretty much everything we're waiting for centres around the popular codec. Shame that we couldn't have cineform or an alternative.
It sounds technically that 4K RAW could be faster to get out of the door, and as Colin says, if there is anything the community can do from our side to lobby Sony then please let us know who/what/where...
thanks
Paul
I personally (IMHO) feel in retrospect that 2K RAW could have been a bit of a diversion in terms of resources and if 4K could have been first then it would have introduced 1080p and 4K solutions upfront and the HFR 2K RAW could have been further down the list.
Taking a straw poll i doubt that many are using HFR day to day (there are some i know). But anyway for whatever real world reasons we're in this timeline now.
Just to clarify however
It sounds like there might be *no* recording from the internal HDMI/SDI output to 1080p compressed? (The comment about the 8 bit solution). Therefore the O7Q wouldn't be a generic substitute for a hyper deck shuttle, recording from an FS100 or other HDMI/SDI cameras?
For me, just being able to record the normal signal would have been a stop gap, enabling me to re-rig and swap the O7Q in for the time being, i'm quite surprised i can't.
Reading between the lines it does seem that perhaps the popular codec might be introducing delays. Having said that i've had no end of problems with DNxHD so i don't see that as an alternative. But pretty much everything we're waiting for centres around the popular codec. Shame that we couldn't have cineform or an alternative.
It sounds technically that 4K RAW could be faster to get out of the door, and as Colin says, if there is anything the community can do from our side to lobby Sony then please let us know who/what/where...
thanks
Paul
Paul- what kind of problems with DNxHD? I typically use 175Mbit 422 10-bit with no issues on Win/Mac. Cineform was never stable with ton of bugs and image artifacts. ProRes has also been solid with no issues. In this age of efficiency and powerful processors, long GOP codecs such as H.264 and even tuned MPEG2 can effectively match the ALL-I codecs at dramatically lower bitrates. 422 10-bit H.264 of any flavor (not just XAVC) would be great to see in the O7Q.
olof
DVXuser Sponsor
XDcam is still the best on the NanoFlash IMHO. 100mb/s is fantastic, even 50mb/s is good. Of course the NanoFlash is only 8 bit but I still use it a lot on the 700 because it works so well and is tiny and light. I use the Odyssey for slomo (2K raw) and monitoring only at this point and it is fantastic for that. I would love to see XDcam on the Odyssey, it is a very efficient codec and CD has lots of experience with it.
I am sure CD will get everything worked out and soon too (this is a relative term). In the mean time we have a great eval monitor and it also makes great WF/Parade scope for your NLE (soon VS as well). I have a "Desk Mount" for my Odyssey bracket and a light stand adapter as well that quickly moves the Odyssey from camera to light stand to desk top. These will be available as accessories in my WEBstore next week.
I am sure CD will get everything worked out and soon too (this is a relative term). In the mean time we have a great eval monitor and it also makes great WF/Parade scope for your NLE (soon VS as well). I have a "Desk Mount" for my Odyssey bracket and a light stand adapter as well that quickly moves the Odyssey from camera to light stand to desk top. These will be available as accessories in my WEBstore next week.
I'm only half joking when I say this, but be careful what you say. Now everyone is going to expect a 10-bit codec on February 15th, which is fine with me. I probably miss-interpreted, but when the original timeline was announced I was only expecting an 8-bit codec by December 31st because I thought that would be the easiest and quickest to get out. But hey, I'm not complaining if you guys are only focusing on 10 bit and better codecs.
nyvz
Veteran
I think you are misunderstanding? I think CD only ever wanted to do a 10bit codec, there is no 8bit codec. 8bit in a 10bit wrapper is what comes out of the fs700 in non-RAW mode. When they deliver 10bit compressed recording before adding compressed modes to fs700's raw output, you'll still just be getting 8bit recorded as 10bit from the fs700's 1080 sdi/hdmi output. You wont be able to get 10bit out of the fs700 until the fs700 raw option gets updated I think.
Last edited:
Hey Noah, thanks for clarifying. That's what I meant. I know that we can only get 10-bit from the RAW stream and that anything else will only be 8-Bit in a 10-Wrapper; and I'm pretty sure that's what the 7Q is getting first. So yeah, on the FS700 we are only getting an 8-Bit codec, even though it's labeled as 10 bit. Like everyone else, I would love a real 10 bit codec.
But at this point I would be happy with even 8 Bit 4:2:2 in a 10 bit wrapper.
Last edited:
paulcurtis
Veteran
If working in 32 bit float then various flavours of DNxHD do not decode correctly if superbrights are present, they generate coloured speckled highlights, i've documented that here somewhere here in the past. This means both After Effects and Nuke do not read them correctly. It's fairly well documented but quite obscure. Avid seem to have no interest in doing anything about it. I think Avid products are fine with it. Also DNxHD is 1080p only. Finally i've had issues with alpha within DNxHD too.
A lot of this is QuickTime windows as well.
Cineform as a codec seems pretty robust but the QuickTime interface is a bit ropey too. The new Nuke has better QT support and there's a couple of hurdles to cross there and i hope that CF will work properly in Nuke soon. AE has no issues with it.
Quality wise cineform seems pretty decent 12 bit 4444 RGB with a lossless alpha compression.
ProRes 4444 is also very good but i'm on windows. If Apple weren't so shortsighted then ProRes would have taken over totally.
I've never seen decent quality AVCHD and also colour depth is limited to 8 bit commonly but i know there is a 10 bit version but support doesn't seem that good.
It is *very* difficult to find a good mezzanine codec that's robust and cross platform. For all its quirks CF seems closest.
I would like a version of EXR with Piz compression (lossless wavelet) in a codec container (reason being is that movie files work faster over the network than tens of thousands of separate files)
cheers
Paul
paulcurtis
Veteran
It would be great if Mitch can confirm.
Thinking sensibly about it i'm sure there will be a 1080p compressed HDMI/SDI recording option. As it stands now the O7Q can't record from an FS100 or any other prosumer camera in anything other than 10 bit dpx files.
Now for us the RAW recording would be nicer and have a higher priority but i presume the reason for not recording the standard output right now is just because of the popular codec support...
cheers
Paul
Mitch Gross
Veteran
I am confirming.
Compressed codec roadmap:
Step 1 -- 422 10-bit 1080 (30p/60i) from 1080 source.
Step 2 -- 422 10-bit 1080 60p from 1080 source.
Step 3 -- RGB 10-bit or 12-bit 444 1080 60p from 1080 source.
Step 4 -- 1080 60p derived from 4K RAW (FS700 & C500)
Step 5 -- 422 10-bit 4K 60p from 4K RAW
Step 6 -- 422 10-bit 2K (?p) from 2K RAW (not sure how fast we can sustain encoding)
The first two steps are absolutely what gets done first. After that the order may change. The frame rate on the last one is unknown right now as it is hard to know how quickly we can encode the compression to maintain the data rate window. 240fps is really fast and the window for data rate has to stay very tight. Have something suddenly change in the image (as is want to happen in high frame rate work) and it could "break" the codec.
Anyway, the first two are the big hurdles. After that much of this work happens concurrently, which is why the order might change. There may very well be an update where a whole pile of them arrive at once. Time will tell.
BTW, for clarity's sake, please note that this info was part of our original Firmware roadmap at the start of this thread. Dates have shifted; goals have not.
morgan_moore
Major Contributor
You mean up to 30p of course?
paulcurtis
Veteran
Brilliant, thanks for the clarification. Just one more question, the 10 bit 4K, will that be log encoded as originally planned? In fact all the 10 bits, can they be log encoded?
I would also like to ask very nicely whether as well as the 1080 from the 4K source, if there could be a straight 2K from the 4K source in the same manner. I'd like to use the full raster from the 4K and a straight half downsample could be done at very high quality. Any idea whether this would be 422 or 444?
cheers
Paul
Last edited:
sunnyjimfilms
Well-known member
Ok, I know we've done this to death and frankly it's giving me a headache but what does it mean - 10 bit 422 from a 1080 source? Is that true 10 bit or 8 bit in a 10 bit wrapper? Surely true 10 bit has to come from the raw source? Confused.