Welcome to our community

Be a part of something great, join today!

  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

The First Komodo Out in the Wild

Aaron love these calculations.
Here's a quick preview of an 8K Komodo Sensor:
8192 * (27.03/6144) = 36.04
Notice it's EXACTLY Vista Vision. Not sure if that's a coincidence or not, but worth noting.

Actually, 36mm in width is 35FF - not VV. Old school VistaVision is 37.72 in width and RED's version of VV is just shy of 41mm wide.

IAC, if the 4.4 micron pixel pitch of Komodo is part of a sensor design rubric for DSMC3, I'd expect closer to 10K (44mm) in width. There have been a number of lenses come to market in the last couple of years that can handle more than 35FF 24x36mm coverage, including RED VV. That said, I do think there is a sweet spot beyond which you're essentially going medium format. Nothing inherently wrong with that, but for most narrative production scenarios having a MF target with it's reduced DoF might not be welcome.

FWIW, assuming Bayer CFA on CMOS, 10-12K to oversample for 8K seems like the obvious next step. Helium sensor tech is just over 3.6 microns, so it would be about 44.85mm wide at 12K and just over 36mm at 10K. RED could easily have sensor tech in the pipeline for DSMC3 at a completely different pixel pitch, only time will reveal that.

What is the "ideal" pixel pitch? Gemini is 6 microns which is part of its superior sensitivity equation, but certainly not the whole story. A 12K chip at 6 microns would be 72mm wide, which would be fun for certain use cases, but unwieldy for most production scenarios. 12K at 3 microns would be 36mm wide - to match 35FF it would need to be 8K tall - 96 megapixels anyone? How about an optical path capable of 165 lp/mm (line pairs per millimeter) across 90% of the frame? Dragon and Monstro are 5 microns which works well with most of the cinema optics available - but 12K would be 60mm wide!

It would seem like the sweet spot is 4 microns or so. 10K would be 40mm wide, which is too much for most S35 lenses but many 35FF optics would cover. 12K would be 48mm wide, which would limit lens options - but still have enough DoF for narrative. Getting wafers that big without dead pixels is a whole other discussion. IMO, a 12K wide sensor should be 6K tall (24mm at 4 microns) for a potential 72 megapixel raster. I know the anamorph peeps would want even more, but 24mm should get them there.

With a sensor that large - 24x48mm - you really open the window, for windowing. How about 8K 16:9 at 32mm wide for sports/docco/verité? Should work with almost all S35 glass and have decent DoF - and, perhaps - decent HFR. That would suit a lot of production scenarios, including 8K shows - yes, actual resolution after demosaicing would be likely be closer to 6.5K, but I'd guess we are more than a decade away from anyone worrying about that. The full chip would be spectacular for some projects and 125 lp/mm is a good match to premium optics.

Cheers - #19
 
I'll make a tidy graphic, but I'm holding off due to to a couple of things that remain to be seen.

Until them format purists, VFX artists, and true believers enjoy some format witchery!

Film Format Relevance to the Komodo 6K sensor:

5.5K 17:9 = Super 35mm 3-perf
3K 16x9 = Super 16mm

About 5K is the with of Academy 35mm format for the 3 of you who still use zooms like that, but the image circle will cover larger than that for sure.

There's more to dive into there, but those are the most notable and of importance in my mind.

Digital, well Komodo is around APS-C, so all that glass.

For those who might go rogue and might adapt Micro Four Thirds lenses to Komodo's RF Mount, likely very safe at 4.5K 16x9. Will depend on the lenses if it covers beyond that.


And highly relevant to those of us working with other RED cameras and using Komodo!

Relevant to RED if matching formats:
RED Monstro and Dragon 5.5K FF w/ a 4% extraction = Komodo 6K 17:9
RED Helium 7.5K FF w/ a 4% extraction = Komodo 6K 17:9
RED Gemini 4.5K w/ a 2% extraction = Komodo 6K 17:9

Obvious resolution disparity and certainly frame rate differences, but this a pure format basis thing.
 
Any word yet about the final choices of resolutions and formats that Komodo will include Phil?
 
Any word yet about the final choices of resolutions and formats that Komodo will include Phil?

I sort of know, but we'll see what really happens.

Main thing I can say is I don't view the "loss" currently of 2:1 as a big deal, despite me loving that format. It is a very minor resolution extraction from the full 17:9 frame IMO. Might net you a 1fps or whtever, but probably not.

Interestingly RED has strayed from the normal FF, WS, and HD nomenclature to a more numerical 17:9, 2.40, and 16:9. I don't hate that. Though 17:9 makes me feel tingly sometimes as I've been a DCI, UHD, and Scope sort of dude.

I'll cry later on that I guess.

Anamorphic is the big question and we know support is coming which is nice. As to how much, how, and which anamorphic support we get; who knows. I think it's safe to say 2X is coming though.

This is all good in Komodo's specific case as it's an exercise simplification and reduction in mind.

Currently having a Komodo these spherical formats:
- 17:9
- 16:9
- 2.4:1

Makes sense to me. We'll know soon on anamorphic.

As for resolutions, well yeah, I want every K and .5K variant in there of those things. Mostly as I mentioned it opens up not only a world of glass possibilities and formats, but also additional frame rates.

Truthfully, it will be a rare day for me to film under 5K resolution on Komodo as I mainly deliver 4K and 8K content professionally. Maybe down to 4K I guess. And I do have the itch to do some 3K Super 16 work. That's a weird itch though.
 
Frame edge to frame edge is 37.72mm, but wasn't the actual extraction area of VistaVison 36mm?

Extraction was variable back in the day all the way to modern VV use.

Even more interesting is we never talk of Arnoldscope, which was 10-perf VV, just a smidge wider.

#runsAndHides
 
Even more interesting is we never talk of Arnoldscope, which was 10-perf VV, just a smidge wider.

Why am I not surprised that such a format existed? Mind you, that's a long length of film. I'm certainly not surprised that nobody uses it!

BTW do the anamorphic modes on Red sensors allow for a higher frame rate?
 
BTW do the anamorphic modes on Red sensors allow for a higher frame rate?

Historically, no. Can't say how that changes with the global shutter, but vertical scan height always made the difference in HFR in the past. So since anamorphic modes kept the same scan height as their respective full width counterparts, we didn't get any benefit. It'll be interesting to see if we get any bonuses this time around with Komodo.
 


I won't bore y'all with the words I posted on instagram as I posted much of my findings here today.

The small recap to note is I've tested over 200 lenses on Komodo and thus ends the mega work of figuring out what works, what doesn't, and what's ideal. I have a few more goodies incoming and some images of lenses not-yet-announced, but primes through zooms the news is mainly good.

I've posted the lens coverage for different formats, but the filmmaker's perspective to me is you can use Komodo for Super 35mm and 4K or 2K delivery or Super 16mm 2K delivery. Interestingly I don't really work much for 2K, but if I was a documentary filmmaker I would investigate this camera as it has a global shutter that is ideal for some of that world whether full 6K Komodo filming is desired or some sub-resolution/sub-format.

Doing my best to tackle questions along the way.

As for anamorphic generally this sensor inspires 2X and 1.3X mainly in my mind. Most people haven't figured out the larger format anamorphic optics explore things like 1.25X, 1.5X, and 1.8X. This is for many reasons, but it's mainly a practical limitation as well as focusing on the strengths of each format.

For those unfamiliar with Anamoric Workflows, traditionally on wider aspect ratios you use a smaller nearly square-ish rectangular crop and then desqueezed the um... squeezed image. For instance on Monstro 8K 6:5 anamorphic 2X unsqueezes to a delightful 2.4:1. Using the full sensor at 2X nets a very extremely wide 3.79:1 ratio, which is madness, but hell this is sparta so kick with the sandals on and have some fun.

Most people do truly think of 2X anamorphic when they "think" anamorphic and fortunately there's at least two fairly affordable options incoming on that front, but I can't really dive into that yet.

1.3X however is interesting on the full sensor with a 2.53:1 ratio coming out of it, more likely practical is what 1.3X was truly designed for 16:9, which comes out to a ratio of 2.37:1.

Hope all of this is useful stuff. I get to play for a little while. Been dying to get back into the world to actually film some stuff. For those not in LA as of today, June 12th 2020, that became a bit easier.

With caution I venture out into the world to film once again. More soon.
 
Thanks for sharing your findings Phil!
1.3X sounds very interesting :)

As for 3.79:1, it's not that far fetched. I've shot 4:1 (crop from the full sensor) for some corporate gigs. So the possibilities are endless!
 
... and fortunately there's at least two fairly affordable options incoming on that front, but I can't really dive into that yet.

Makes me almost forget about how much I'm looking forward to Komodo!

If it's a manufacturer one could have heard of, my guesses would be either Schneider oder Tokina. Discussed this with a rental company recently and we couldn't quite believe that Schneider (ISCO), a company that was one of the biggest names in anamorphic did nothing on that front in the last - I don't know - 30 years? (Of course they are not Panavision or Vantage/Hawk, but compared to Zeiss e.g. who never had much to do with Anamorphic before they released the Master Anamorphics, except the Ultrascopes in the 50s - Schneider always was a company built closely around the anamorphic topic) They built actual anamorphics for Arri, pretty much most of the projection lenses used for anamorphic projection so I couldn't see why they would actually round their Line-Up (that is arguably under rated, but most people don't even look at Schneider lenses or know that there are some - the Xenar III are really stunning, Xenons maybe not so much) with an anamorphic set. Tokina could be another guess because I think when they released the Vistas they meant serious business and one set of cine primes is probably pretty litte to justify on it's own over years for a manufacturer...
 
framerate workarounds

framerate workarounds

Hey Phil!

I was wondering something that RED doesn't really tout. On a previous epic w i worked on and some dragons, you can go 4:1 and 8:1 and get some really high frame rates in high resolutions. Can you get the 4:1 and 8:1 aspect ratios on Komodo, And when if you can, are you able to ramp up the frame rates significantly, since you're using less of the sensor, like dsmc1 and 2 can achieve?

Many don't know nut im pretty sure i was getting close to 200fps in 6k 4:1 and close to 300fps in 4k 4:1.
 
Phil-- looking forward to seeing stuff as you continue to contribute. When you get a moment, though, I'd love to have your opinion about the compression options. We know that you're happy with the quality of the footage, but do you like working with the "HQ" and "MQ" versus choosing your compression ratio?

I'm assuming that the move is meant to reduce complexity, especially for an audience new to Red: much easier to think about "medium" and "high" quality, make it more like a flavor of ProRes. After all, I myself found it very intimidating the first time I shot with Red-- how to pick the right ratio for my project from a dozen options? (Belated thanks for the data rate tools on your website, by the way!) But, after getting a feel for it, I like the ability to dial this in.

(It's also that I like the idea of matching B-cam compression to A-cam-- though that's probably more OCD than actually important. Brief sorta conspiracy theory question: there are hints that the r3d compression on Komodo might be a slightly different beast than DSMC2's r3d. So would 6:1 on Komodo mean the same as 6:1 on Helium, or are they apples and oranges? Perhaps the different naming convention is a way of discouraging false equivalence?)
 
What if you mounted an anamorphic lens rotated 90° to squeeze the horizontal dimension to fit a skinny raster capable of higher frame rates. For example - 3.56:1 using 2X to deliver 1.78:1 after the de-squeeze. Super slo-mo and trippy bokeh - I feel a dream sequence coming on...

Cheers - #19
 
What if you mounted an anamorphic lens rotated 90° to squeeze the horizontal dimension to fit a skinny raster capable of higher frame rates. For example - 3.56:1 using 2X to deliver 1.78:1 after the de-squeeze. Super slo-mo and trippy bokeh - I feel a dream sequence coming on...

Cheers - #19

I wanna shoot 1-perf 35mm with an anamorphic lens turned 90 so I can shoot for 44 minutes per 1000' mag. :)

For an idea of how that bokeh might look:

https://vimeo.com/61184180
https://vimeo.com/316047218
 
Hey Phil, sorry if you have already mentioned it and I missed it but what riser are you using on your komodo? Is it a preproduction of something or is it available now? Thanks for the constant updates!
 
Hey mate, yeah I just scaled up it in an 8K timeline. There’s other more official ways to do it but I just scaled it and it looked fine.

Heh. Don't ask me how many times I've scaled footage up in 20 years. Though I experiment with various processing methods the answer is pretty close to zero times.

I did some AI based scaling work a few months ago though. Pretty interesting. Not likely what people are going for or expecting. Will help get dated and older content up to snuff for 4K and 8K however.


4:1 and 8:1 and get some really high frame rates in high resolutions. Can you get the 4:1 and 8:1 aspect ratios on Komodo,

Unsure if those special formats will come to Komodo as they are really off the beaten path. I'm 90% sure those exist due to military and scientific clients RED has for high frame rate analysis.

Komodo is a tighter ecosystem, but as to what formats end up in there, hard to tell.

Overall though, 2:1, though I love it, it's currently not in there and I don't know if it's fully "needed" as you can extract it so easily from the 17:9 DCI format sensor size.


Fantastic information as always, Phil. You are such an asset to this community. We are so lucky to have someone like you around here. Thank you for being you.

Cheers Dorian, appreciate that. Some would likely argue I merely an ass to the community, but I do my best :)


What if you mounted an anamorphic lens rotated 90° to squeeze the horizontal dimension to fit a skinny raster capable of higher frame rates. For example - 3.56:1 using 2X to deliver 1.78:1 after the de-squeeze. Super slo-mo and trippy bokeh - I feel a dream sequence coming on...

It's a cool and unique look if you can get glass to cover it, but be careful for streaking flares. Most people like horizontal and even diagonal, but vertical flare seems to remind people of good old crappy CCD sensors :)


but do you like working with the "HQ" and "MQ" versus choosing your compression ratio?

It works much the same way, you can choose to go with MQ for instance. If you are in HQ and need to go to a higher frame rate, it will automagically select MQ as well. Personally I'm truly not bothered by the move to HQ, MQ, and LQ. You can still select what you want, but it also is easier on the mind to enforce a more common workflow of just going with the higher quality/highest data rate possible. In the world of DSMC2 I pretty much shoot 2:1 through 12:1, but really what that actually means is I'm using the lowest compression ratio possible at any given frame rate and format/resolution.


Here was my early morning journal thingy.




I'm mainly using 1TB Angelbird CFast 2.0 cards, but I also film long takes and long form content often and those get me about 1 hour of roll time at 6K 17:9 REDCODE RAW HQ, about 10 more minutes in MQ, LQ as of yet, unknown. I think many will live in the 256GB and 512GB range for day to day filming. I personally wouldn't skimp on media and don't use things like cheaper 128GB and don't use 64GB at all as it doesn't have the data rates to support recording REDCODE RAW or ProRes. There's about 4-6 brands I'd say worth exploring.

The media bay has a removable hatch that reminds me of a miniature truck tailgate and it is indeed removable with a small spring loaded latch. Firstly, I'm super into a protected and contained media bay. But there's certainly some interesting opportunities to feed data out via that CFast 2.0 connection to other drives for higher capacity rolling or perhaps direct to disk workflows. I imagine I'll tinker with that later, but the potenital is there for some fun workflows.

You can see I have mics and all the goodies piped in. Komodo in my opinion is indeed quiet enough for use of an onboard mic. Ideally driven with a good preamp or something that can boost itself. Komodo has preamps and you can adjust the gain, but for lowest signal noise you want to feed it a good quality low noise signal. Traditional sound on set hasn't been a huge deal if you know what you are doing with DSMC2, but Komodo has opened up this door again and I think it's alluring for hot and fast documentary-style work for sure.

Doing weird stuff with Komodo this week. More footage likely at the end of the week once I'm done jogging around.
 
Back
Top