Odyssey7 & Odyssey7Q Status Update

Some update here guys I placed an order with Adorama the 16th of October, this morning I received an email saying that my Credit Card was charged meaning my order will be dispatched soon :)
to Clarified I ordered the 7Q and the two 256GB Hard drives
 
Some update here guys I placed an order with Adorama the 16th of October, this morning I received an email saying that my Credit Card was charged meaning my order will be dispatched soon :)
to Clarified I ordered the 7Q and the two 256GB Hard drives
I was afraid of stuff like this... I order 3.5 months earlier and might get it later. Bummer.
 
It's worth a look at certainly, thanks for the explanation and i won't ditch the shuttle just yet!

When you say you plan on tweaking, would the higher quality work mean scaling a debayered 4K image down or still using the same, but improved, binning techniques - can the 7Q downsample that 4K stream in real time? I would have thought (but could be wrong) a downsampled debayered image would be higher quality that a binned image?

thanks
Paul

Well spotted Paul! Reading Mitch's reply It does sound like the '4K derived' is using the same 'Pixel binning' method as the 2K Raw read-out, or something very similar. If this was the case you would expect the same aliasing issues as are already known about in the 2K Raw footage, due to the OLPF being optimised for 4K - in fact it's hard to see what benefit you'd get over shooting 2K Raw - apart from being able to avoid transcoding.

Previously I was under the impression that the HD was being derived from 4K so as to avoid these aliasing issues - so maybe we've misunderstood what Mitch was saying. Maybe the binning is only for the monitoring output?

Mitch: if you get a moment could you please clarify this (although I understand your probably massively busy right now!)

Thanks,

Colin Elves
 
Mitch, Will the 7Q in the future be able to crop a 1:1 1080p crop image from the 4K sensor?

Since it downsample the 4K, this should be difficult to achieve...

I do wildlife and the 1:1 crop reach and 240 fps would be amazing, actually I don´t have a fs700, I use a RED ONE, that makes the 1:1 crop from 4K sensor and 120fps compressed RAW, If you put this feature in the 7Q, I will sell my RED ONE and buy an fs700+7Q
 
This is not the same thing as binning, this is compiling 4 photosites of three different colors to arrive at one pixel. Binning takes photosites and groups them together to create fewer "virtual photosites". There are drawbacks and advantages to the technique, but again I will point out that it is pretty much exactly what Canon does on its Cinema EOS cameras quite effectively.
 
Mitch, Will the 7Q in the future be able to crop a 1:1 1080p crop image from the 4K sensor?

Since it downsample the 4K, this should be difficult to achieve...

I do wildlife and the 1:1 crop reach and 240 fps would be amazing, actually I don´t have a fs700, I use a RED ONE, that makes the 1:1 crop from 4K sensor and 120fps compressed RAW, If you put this feature in the 7Q, I will sell my RED ONE and buy an fs700+7Q
You're asking a recorder to do something that a camera needs to do. We would have to dismantle each frame to extract just the center and then recompile the data. The FS700 is only 240fps capable in 2K RAW. If we were to take the 4K RAW out of the camera and then somehow extract the center 1080p, we would still be limited to the frame rate the camera is capable of at 4K.
 
This is not the same thing as binning, this is compiling 4 photosites of three different colors to arrive at one pixel. Binning takes photosites and groups them together to create fewer "virtual photosites". There are drawbacks and advantages to the technique, but again I will point out that it is pretty much exactly what Canon does on its Cinema EOS cameras quite effectively.

Ah that makes sense, though I guess it's not totally clear if FS700 2K is from binning either. Makes sense that if it is binning it would have more aliasing issues since binning is combining same-color photosites that are twice as far away from each other as other-colored photosites which are immediately adjacent and used in the RGB super-pixel mode. Then again I'm still not sure it makes sense that 2K is binning anyway, since that doesn't really explain the increased noise, perhaps it is skipping...

One concern is that if 2K/1080p from 4K is using RGB superpixel "debayer/downscaling" but 2K/1080p from 2K HFR is using a more traditional debayer, they may look quite different.

I wonder what kind of OLPF Canon has in the C300/100 and whether they are optimized for 1080p or for 4K like the one in the C500... And does the C500 scale 1080p from 4K or does it do the C300 superpixel thing for 1080p?

Frankly it might be nice to have a mode that almost perfectly matches C300s (using RGB superpixels and a LUT that approximates C-Log) but records in prores and has a bit more DR :)
 
Last edited:
This is not the same thing as binning, this is compiling 4 photosites of three different colors to arrive at one pixel. Binning takes photosites and groups them together to create fewer "virtual photosites". There are drawbacks and advantages to the technique, but again I will point out that it is pretty much exactly what Canon does on its Cinema EOS cameras quite effectively.

Okay - I think I get the difference. But my concern is that the frequency of the spacing of each 'pixel' will be wider than the minimum frequency passed by the 4K optimised OLPF - so would still suffer from increased aliasing.

Although the C300 and C100 work in the same principle Canon could (would) use an OLPF optimised for HD (and I think the C300 had aliasing issues at the start anyway, didn't it).

Unless you're doing some fancy maths work to reduce the aliasing, or I'm totally wrong - both of which is probable!

It'd be good to know I am wrong about this though!
 
Yeah, no question, it will definitely increase aliasing compared to a proper debayer, but I guess the idea is that there is precedence for that method so maybe it will yield acceptable, if not ideal results. Frankly people have been shooting with line-skipped DSLRs so long and I see so many poorly scaled videos online (a lot of web video players use nearest neighbor scaling for embedded content) that a little aliasing seems hardly noticeable and in some cases almost preferable over soft footage. The saddest thing is when I see soft footage that is also aliased, DSLRs usually...

One thing I'm kind of curious about is whether its actually possible to deconstruct and debayer 1080p RGB image from superpixels back to 4KHD, since there is potentially 75% of the original pixel data there (plus gamma correction and probably bit depth downsampling).
 
For all of you eager beavers who are hounding our poor dealers, please understand that while we are in fact shipping on the Odyssey7Q, it's not like we have unleashed 10,000 units all at once or that there are only a couple places selling them. We have dealers all over the world and we're trying to be fair in sharing the wealth between them. Also, we have a rigorous QC system, which means that every component gets tested and then the entire assembled device gets an intense burn-in test. That means we can only ever manufacture them just so fast.

As we've stated in the past, we are ramping up and have increased our capacity, but it will still take us well into December to satisfy all of the current orders we have for the Odyssey7Q. We also expect that now that images and reports of the Odyssey7Q are beginning to trickle out our orders will increase.

Be patient, my friends. They are coming soon.
 
This is not the same thing as binning, this is compiling 4 photosites of three different colors to arrive at one pixel. Binning takes photosites and groups them together to create fewer "virtual photosites". There are drawbacks and advantages to the technique, but again I will point out that it is pretty much exactly what Canon does on its Cinema EOS cameras quite effectively.

Canon had quite a few aliasing issues at the beginning too :)

Mitch, not sure if you can answer this quite yet as it may be too early, but the proposed high quality 1080p output promised in a future update : Is that from a scaled debayered 4K image or from this technique of compiling 4 photosites?

And just to make sure, this photo site combination happens pre-debayer? Is that right?

My only observation at this early stage is simply that the 700 has a 4K sensor. The OLPF is for 4K. The debayer *needs* to happen at 4K because the reconstruction process needs the neighbouring pixels to reconstruct properly (depending on the debayer being used). Any method of technique to decimate the original signal will never produce as good as an image as a 4K debayer -> downsample or crop.

If people wanted a 1080p camera pick up an F3 - awesome image quality for the price...

(trying not to sound like stuck record, stuck record, stuck record....)

cheers :)
Paul
 
As we've stated in the past, we are ramping up and have increased our capacity, but it will still take us well into December to satisfy all of the current orders we have for the Odyssey7Q. We also expect that now that images and reports of the Odyssey7Q are beginning to trickle out our orders will increase.

Be patient, my friends. They are coming soon.

Hi Mitch - a quick question regarding the 'vanilla' Odyssey7 - is there any news or ETA on this version?

Thanks :)
 
For all of you eager beavers who are hounding our poor dealers, please understand that while we are in fact shipping on the Odyssey7Q, it's not like we have unleashed 10,000 units all at once or that there are only a couple places selling them. We have dealers all over the world and we're trying to be fair in sharing the wealth between them. Also, we have a rigorous QC system, which means that every component gets tested and then the entire assembled device gets an intense burn-in test. That means we can only ever manufacture them just so fast.

As we've stated in the past, we are ramping up and have increased our capacity, but it will still take us well into December to satisfy all of the current orders we have for the Odyssey7Q. We also expect that now that images and reports of the Odyssey7Q are beginning to trickle out our orders will increase.

Be patient, my friends. They are coming soon.

In juxtaposition to previous yang comments I've posted, I'd like to balance them with a little yin. My past experience with C-D, from Flash XDR thru Nanoflash, has been pretty positive. At one point, I experienced a failure of the display on my Flash XDR. C-D replaced the screen free of charge. Mike, Tommy, Dan, and now Mitch, really run a laudable operation there in denver. One of these days I, simply, must stop in. But, my point is to just comment on how admirable their commitment to quality and customer satisfaction is. Great job, guys! You have my faith and support.
 
Canon had quite a few aliasing issues at the beginning too :)

Mitch, not sure if you can answer this quite yet as it may be too early, but the proposed high quality 1080p output promised in a future update : Is that from a scaled debayered 4K image or from this technique of compiling 4 photosites?

And just to make sure, this photo site combination happens pre-debayer? Is that right?

My only observation at this early stage is simply that the 700 has a 4K sensor. The OLPF is for 4K. The debayer *needs* to happen at 4K because the reconstruction process needs the neighbouring pixels to reconstruct properly (depending on the debayer being used). Any method of technique to decimate the original signal will never produce as good as an image as a 4K debayer -> downsample or crop.

If people wanted a 1080p camera pick up an F3 - awesome image quality for the price...

(trying not to sound like stuck record, stuck record, stuck record....)

cheers :)
Paul
Can't answer for what our future "high quality HD" system will be. To be clear, the current monitoring system does not deBayer at all, which is the same as on the Canon cameras.
 
Hi Mitch - a quick question regarding the 'vanilla' Odyssey7 - is there any news or ETA on this version?

Thanks :)
It is scheduled to ship in January, but if we get ahead of certain things we'll try to push it out sometime in December. Much of the work for the Odyssey7 is encompassed within that of the 7Q.
 
Back
Top