KOMODO....

[Bastien Tribalat]

Where did you get the 12.8 at 2k?

The C5D review has the 11.8 and even less at 3200.

I would agree that Komodo, in this day and age, should be aiming for a lot more than this, even as a B Cam to a DSMC2.

cheers
Paul

Agreed.

Question for the sake of it (for anyone here) : would you rather have more stops of DR in the shadows or the highlights ? Like if you could choose (but don't be a wise a**, you are not allowed to answer both)

ME : if Komodo has like 14 true stops of DR, I would find useful to have a bit more DR in the highlights.

 

[Christoffer Glans]

Agreed.

Question for the sake of it (for anyone here) : would you rather have more stops of DR in the shadows or the highlights ? Like if you could choose (but don't be a wise a**, you are not allowed to answer both)

ME : if Komodo has like 14 true stops of DR, I would find useful to have a bit more DR in the highlights.

Definitely more in high-lights, but also I rather have a roll-off in the highlights so that it never clips sharply. Both Red and Arri does this great, Red didn't have it with the MX, but introduced it on the Dragon. But BM and most others don't have this, they use a much more linear stepping of each stop, meaning as soon as you hit top highlight, it clips sharply. So you need to always expose for the highlights and you can never do "happy accidents" with other cameras that doesn't use a roll-off. When I shoot with my 6K I can over-expose and never be worried that it'll clip badly. Highlights will not be able to be recovered, but they don't look bad. So working with a Pocket 6K requires much more careful exposure, I'd say working with cameras like the Pocket is much harder than a Red, Sony (Venice) or Arri.

That said, if you don't have usable stops at the lower end, you'll end up with underexposed footage that you can roll off highlights in grade, but that looks like shit because of the overall lower values. So there has to be a balance between them; a nice roll-off, but at least two-three stops of usable dark shadows, clipping shadows looks as bad as clipping highlights.

 

[Misha Engel]

Where did you get the 12.8 at 2k?

http://www.reduser.net/forum/showthread.php?177459-KOMODO&p=1867364&viewfull=1#post1867364

The C5D review has the 11.8 and even less at 3200.

I would agree that Komodo, in this day and age, should be aiming for a lot more than this, even as a B Cam to a DSMC2.

cheers
Paul

The real problem for the Komodo is that it's kind of late to the party and people start buying alternatives.

 

[Christoffer Glans]

http://www.reduser.net/forum/showthread.php?177459-KOMODO&p=1867364&viewfull=1#post1867364
The real problem for the Komodo is that it's kind of late to the party and people start buying alternatives.

Being late to the party doesn't have to be a bad thing and if Red is as fast as they usually are with turnarounds and changes, they would take the alternatives seriously and make changes to have Komodo be the one camera that disrupts the alternatives, making people question why they didn't wait for the Komodo. But if not, being late, with the same specs as others, is unwise. Better to then be even "later" to the party and really suckerpunch the competition into oblivion.

 

[jordan levy]

^ yep, currently where I'm at with it. But really it's going to come down to general specs once we get them.

Since this is a cubic cam people will need to accessorize it accordingly. I actually think the battery solution is alarmingly clever likely making this really ideal for run and gun or quick setups when in a pinch a Director might yell "hey, can we get a camera on the ceiling?!". Rigged properly this could be a doc filmmaker's dream too. I don't know Soderbergh (I did meet him while working on Solaris at Rhythm and Hues in the early 2000s, we did the wispy stuff which actually benefited Riddick later on), but filmmakers like him have been really trying to discover a small camera that helps them work wicked fast. I personally think DSMC2 is pretty damn small as is if you rig it right, but Komodo is significantly smaller and apparently a lower power draw.

My only wiggle is the 3.5mm jack for audio, but that's not the end of the world at all. Hell that might even be how we deal with timecode even if the EXT port doesn't handle it. There's clearly no space for a full size XLR and from the looks of it not much for one of the smaller connectors where that port is. I'm just trying to figure out how to rig it in 20 different ways for 20 different types of productions.

I see potential for riser accessories that do things, like a sneaky two phantom power XLR box under there that securely mounts might be pretty damn cool and it would help with the common rod height spacing for a follow focus.

I'm also curious about the accessory mounting outside of what we know. Jarred's clearly not showing us the top of the camera for some reason. I suspect bolt hole cheese, connectors, buttons, or some sort of combination of all that. I think the weirdest thing would be an integrated menu screen there w/ monitoring perhaps, but who knows. That would be really interesting to have some sort of fold out and spin LCD touch screen there, but that's not a lot of space up top. One of my early asks, though not possible, was to get video out of the DSMC2 Sidekick because I was working with drones early on a bunch and it would have been a nice sanity check when wireless dropped due to interference in big cities.

Since RED is integrating so much here; mount and battery, it wouldn't shock me if they could pull off an integrated articulated display up top. It would be impressive as hell actually considering the real estate available from everything we've seen so far. It would need to be a robust mechanism because in the world of smaller mirrorless cameras that is very much something that breaks early and often for some shooters. But maybe there's more to it that that, maybe something else is up there. I'm shooting in the dark here until I see it in person.

If it is fully integrated and all you need is a lens, media, and battery to film something; this would be the first time RED has done anything so streamlined. Even without the LCD, it is already.

That would be the best thing that could happen now, is if it had a screen at least as big as a dslr for settings and viewing so you didnt need a monitor rigged up if you were using it for something like you said; drones, stage rigs, gimbals, etc. this would make a massive difference in my consideration to buy it.

 

[Fahnon Bennett]

They did a review on the Canon C300 MKII where they could clearly see that Canon's measurement of 15 stops was clearly false advertisement since almost five stops at the low end were totally unusable, so the C300 MKII is essentially only a 10-10,5 stop camera.

I think there's a bit of a misunderstanding on this stat. As far as I know Cinema5D never found the DR of the Canon to be 10-10.5 stops, but did find it to be 12.3 stops. They also found the FS7 to be the same, and typically find cameras to be lower than other sources, with the exception of the Alexa, which they rate at 14 stops. Here is the original article:

https://www.cinema5d.com/canon-measured-15-stops-dynamic-range-c300-mark-ii/

"In our tests we rated the Canon C300 mark II at 12.3 stops, about 2 stops weaker than the Arri ALEXA that is rated by us and by the manufacturer at about 14 stops".

Their test also isn't the last word, as others like the EBU have measured and found the same as Canon (15 stops). Source:
https://www.newsshooter.com/2016/02...ebu-tier-1-standard-for-hd-and-tier-2-for-4k/

"His key finding is that the camera has about 15 stops of dynamic range in Canon Log2. His method is based on standardised testing of the signal-to-noise ratio within patches of a colour chart under tungsten light".

Being a former MX owner, and a current C200 owner (maybe future Komodo owner), I can tell you the Canon sensor is higher DR than the 13-stop-rated MX sensor was if you're shooting in C-Log 2 (not 3).

 

[David Rasberry]

To me, that's what counts the most because as an editor and colorist I hate when people on set say "don't worry, we deal with it in post".
Sure, there might be occasions where you really don't have a choice but otherwise it should always be correctly exposed so you have the maximum possible DR (not even on the entire frame but on what matters to the director) and less grain/noise.
(I think denoiser tools should not be there as part of a workflow but as tools to deal with mistakes)

Well said. I would add that there is really no good reason to underexpose modern digital cinema cameras during capture.

Cheers - #19

Yeah, far too many less knowledgeable shooters take uber high ISO settings seriously and wind up with colorless lifeless noisy mud.

That said, every video camera incorporates some degree of noise reduction into its encoding process. Raw cameras don't. The judicious application of modest noise reduction profiles to raw footage is a normal part of RAW workflow IMO.

 

[Paramon Ivanovich]

being late, with the same specs as others, is unwise. Better to then be even "later" to the party and really suckerpunch the competition into oblivion.

I hope for sucker punch, i really do,
but how are you going to sucker punch S35 6K raw for $2500? i don't think it's possible.

 

[Gregory Karydis]

The real problem for the Komodo is that it's kind of late to the party and people start buying alternatives.

Then people should start making back the cost of these investments soon!
Because once Komodo is out, the market will be flooded with all of these stop-gap and "tide-me-over" solutions they bought and everyone will want to get a Komodo.

Being late to the party doesn't have to be a bad thing and if Red is as fast as they usually are with turnarounds and changes, they would take the alternatives seriously and make changes to have Komodo be the one camera that disrupts the alternatives, making people question why they didn't wait for the Komodo. But if not, being late, with the same specs as others, is unwise. Better to then be even "later" to the party and really suckerpunch the competition into oblivion.

Exactly!


I am confident Jarred has a whole deck of aces up his sleeve!

 

[Bastien Tribalat]

I hope for sucker punch, i really do,
but how are you going to sucker punch S35 6K raw for $2500? i don't think it's possible.

Speaking of sucker punching, and I know it is going off topic but anyone know if DSMC3 will have a 70mm-ish sized sensor in its roster ?

 

[Blair S. Paulsen]

It is my understanding that making large, high resolution sensors is an expensive proposition due to poor yield ratios per wafer. Per Jarred's posts, RED's Dragon VV program was hammered by the percentage of rejected chips. I think it's in RED's DNA to keep pushing for large format sensors and I expect they will eventually make a 70mm-ish cut. FWIW, like Phil Holland, I consider VV to be the Goldilocks target size for capturing "human scale" content.

For big landscapes and some other specialty use cases, I'll take the biggest sensor I can find quality lensing for - but I'd expect it to be a few more years before the electronics to support even 24fps on a >100mm, 32K sensor will be possible without NASA level cooling tech and another generation of silicon development. Yes, the military may already be there, but for obvious reasons I doubt they'll let that kind of tech into the wild anytime soon.

Cheers - #19

 

[Rakesh Malik]

For big landscapes and some other specialty use cases, I'll take the biggest sensor I can find quality lensing for - but I'd expect it to be a few more years before the electronics to support even 24fps on a >100mm, 32K sensor will be possible without NASA level cooling tech and another generation of silicon development.

When shooting landscapes with my Red, I do miss my 4x5... but good luck capturing motion with a view camera!

Yes, the military may already be there, but for obvious reasons I doubt they'll let that kind of tech into the wild anytime soon.

Cheers - #19

I think the days of our military leading the way in R&D are long past.

 

[Brian Boyer]

Question for the sake of it (for anyone here) : would you rather have more stops of DR in the shadows or the highlights ? Like if you could choose (but don't be a wise a**, you are not allowed to answer both)

ME : if Komodo has like 14 true stops of DR, I would find useful to have a bit more DR in the highlights.

Highlights.

After focus, I believe the two technical factors that contribute the most to the perception of higher quality images are color and how the highlights are handled.

Obviously, color plays no role in b&w photography, so you're left with highlight handling as the most often used metric of technical image refinement, whether it's a conscious determination or not.

Beyond a certain point, high resolution isn't necessary to produce a superior image. I can also think of scenarios where a high contrast ratio isn't, either - a Caucasian skier with blonde hair, wearing a white snow suit, on a snow-covered mountainside with a bright, daytime sky as a backdrop, for instance.

To be clear, I'm making a distinction between good image quality and high image quality. You can get good quality images from almost every camera ever made under the right circumstances.

I'm not talking about being able to make out detail in a specular highlight on a chrome surface. I'm saying the closer a sensor can perform to how our eyes see bright areas in real life, the more pleasing the images that sensor captures will appear to us.

Arri figured as much (it's in their white paper) with the Alexa/ALEV III after making almost no splashes with the D-20 and D-21.

Why do I discount noise as a determining factor for high image quality in most instances?

Film was never a great performer in low light and it was the gold standard in image quality for over 100 years. Humans are extremely tolerant of noise - film grain, dust and scratches, clicks and pops, wow and flutter, tape hiss, snow, static, etc. You name it and we'll pretty much tune it out. The brain is very good at that.

We're conditioned to pay more attention to where the light falls. This is true in everyday life - walk into a dark room and turn a light on. Need more light in a localized area? Turn on a lamp or a flashlight. You don't curse your eyes and gouge them out because you can't see in the dark. You illuminate the areas you want to see.

It's also true for how movies have traditionally been lit. Noir relies heavily on directing our eyes to only certain areas of the lit frame and on our ability to ignore the dark areas with ease.

That's why scares are easy to come by when something pops out of the shadows. We're not paying them nearly as much, if any, attention. Brighter areas generally garner more attention.

While you may have dark imagery containing no particularly bright areas, I'd say it's much harder to judge quality in a dark image absent our most commonly used cues, and perhaps only then will people look to noise in the shadow areas as a cue in determining a more refined image.

It seems to me that the greater the light intensity a sensor can capture, the more positive effects it would have on the image as a whole.

So, I don't know if it makes sense in terms of the physics or science of sensor design, but if RED produced a camera that captured 7.5 stops under/9 stops over (a DR within their current claims for all their sensors) 18% gray @ ISO 500, it would hit a sweet spot that would please a whole mess of folks.

It would also more than likely further increase their standing and adoption among their primary target audience, methinks.

 

[Savannah Miller]

Also one thing to keep in mind, is even though Komodo is around the corner, we don't know what Blackmagic is working on. The IMX410 (the sensor in Z-Cam F6, Kinefinity Mavo LF, A7III, S1H, Sigma fp, Nikon Z6, is a wildly popular FF sensor with great dynamic range and image quality. I would not be suprpised if Blackmagic is working on their own camera with this sensor and will easily rival Komodo due to better lowlight performance and dual-native ISO capabilities, beyond even what the Pocket 6K does.

Blackmagic getting around RED patents with partially-debayered raw has led Z-Cam to copy and introduce ZRAW.

If RED patents don't hold up for compressed raw capture, we could likely see more cameras offering internal RAW video as it does generally seem that a lot of camera manufacturers seem to gradually increase what they are willing to put inside their cameras.

 

[Phil Holland]

The IMX410

I have no urge to knock Sony cameras at all, in fact I own a few, and they certainly make good business out of selling lots of sensors. But something that is interesting about this particular sensor and the growing knowledge base from likely slightly more tuned in individuals recognizing that several cameras are using the very same tech is the concept of variety. It's only wildly popular because it's a sensor that other companies can put into their systems and provide something that competes nicely in the current landscape, but the issue is there that it's all the "same". You will get a variety of different levels of performance based on exactly what each individual manufacturer deploys in terms of image processing, codecs, general electronics, workflow, etc. But the fundamental concept of one sensor so broadly stretching across systems is actually new here, it's certainly more than I can recall prior to this moment. At the same time all of these manufacturers have more reasons than ever to get back into the game right now. The general consumer doesn't really care about any of this and likely doesn't think of this much. Which is probably for the better as they can focus on which camera best fits them in terms of usability and budget.

Many people who are fond of RED, ARRI, Canon, Sony, whoever feel confident in the quality of the image and some of the uniqueness that comes from that individual ecosystem. RED couldn't thrive if they were merely matching the ALEV III or vice versa. The growing concern from the Venice 6K FF crowd is apparent in my market while across the nation a new TV show just started filming on the Venice w/ 3 cams. Most of what's filming around me is RED or ARRI, but that's mainly the areas I deal with. Though core technology limits of the A9 were to downsample from the 6K 24p for a 4K image which was developed in 2016, released in 2017 in two forms, and evolved further into the sensor many now are just getting.

But the glaring thing here for me is Sony and Canon being oddly quiet, likely not for much longer as a few new cameras targeted at a few different tiers between this month all the way through the middle of next year. Both of them have demonstrated much more interesting sensor technology that they'll likely keep exclusive for a couple years at least, Canon in particular. We just haven't seen them pop up in cameras yet. Video is such a focus today I fear that many of the reasons we're not seeing the crazy high resolution still cameras is it's potential impact on video recording. Right now many are scratching their heads about the A7rIV, which I'm likely buying manly for stills, and it's crop factor towards 4K video. The real high res stuff just doesn't have the readout to do motion really at those levels.

That's all off topic. I do think Komodo will stand on it's own and be unique, that's sort of RED's thing whether you love them or hate them (p.s we know the three in this thread who fall on the latter). Though I enjoy the innovation of modularity and upgrade-ability, I'm an image quality and features first sort of fellow. I've mentioned here if somebody made a camera that looked like a watermelon with a lens stuck in it and I liked the images off it, I'd happily find a way to shoot with it. Which is hilarious to me in retrospect because while filming in San Fran after saying that years back and looking at the Shotover F1, it's round with a tubular protrusion, and it does exactly what I need it to do and it's very much the best at what it does.

The only thing we know so far to the strengths of Komodo are it's form factor, general resolution, and integrated components like battery mount, lens mount, SDI out. That's already new ground that RED haven't been in before. I hope soon, perhaps around or after IBC, we'll get the full story of what's going on.

 

[Bastien Tribalat]

Also one thing to keep in mind, is even though Komodo is around the corner, we don't know what Blackmagic is working on. The IMX410 (the sensor in Z-Cam F6, Kinefinity Mavo LF, A7III, S1H, Sigma fp, Nikon Z6, is a wildly popular FF sensor with great dynamic range and image quality.

It's also the sensor that's going in the rumored successor of the FS7 ("FX9", allegedly full frame 6K dual ISO...). Which brings me to the following affirmation : the sensor alone is not the only factor in the final image quality (duuuuuh, I know).
To me, Mavo LF produces way better looking footage that A7III. (I edited a piece shot on Canon C300MKII and Sony A7III and the Sony does not produce pleasing images to me. It's way to "video-esque" for my taste, too noisy, even at base ISO, too harsh. I had to create LUTs using Film Convert to help match the two cameras in the color grading process)

SO if the Komodo Sensor is "just" a bigger version of the Raven sensor (it was slightly sub-S35 @4.5K so IF Komodo is FF or slightly bigger than s35 @6K it seems doable to me) with IPP2, to me it would be more interesting than anything with the IMX410 sensor (except for Mavo LF and the Lumix S1H. I kinda like those two I confess. Given the choice I would go RED tho, but still)

 

[AndreeMarkefors]

Question for the sake of it (for anyone here) : would you rather have more stops of DR in the shadows or the highlights ?

Definitely more in high-lights, but also I rather have a roll-off in the highlights so that it never clips sharply.

Highlights.

So what do you guys even mean by this?

The sensor has a noise floor. Once enough photons trigger a signal above that you have something. After that, you'll need twice the signal to gain a stop, all the way up to FWS.

This is all you'll ever capture—a linear signal response to incoming light. This is the data, this is what's captured. There is no roll off here.

Now you're free to go into your software of choice and apply any OETF or log curve you like. "Change the ISO" too. You can take that captured linear data and push and pull it along a logarithmic function all day long—for presentation. Clearly, this will change how much information is revealed in the shadows (along with noise), or how much contrast/tonalily/compression you want in the highlights.

But this is our creative choice, right? What am I missing?

 

[Christoffer Glans]

Many people who are fond of RED, ARRI, Canon, Sony, whoever feel confident in the quality of the image and some of the uniqueness that comes from that individual ecosystem. RED couldn't thrive if they were merely matching the ALEV III or vice versa.

What makes Red, Red, is their resolution for downsampling, superior compressed RAW and many stops coupled with IPP2 color science. That's the "special sauce" we get from Red. What is happening now is that many others get into the game on the resolution side, the compressed RAW side, but they still lack stops, widely adopted R3D workflow and color science that rivals IPP2. I've never seen Red perform better against, in my opinion, the superior color science of Arri, but IPP2 is so close to Arri's quality that I wouldn't want to lose the rest within Red's sauce just to get that color science.

The problem, however, is if the competition improves upon all areas that make up the Red sauce. ZCam releases an 8K FF camera, but they lack in framerates, smooth compressed RAW workflows, have a high rolling shutter etc. Blackmagic releases a Pocket 6K with some heavy punches, but also have a much smaller s35mm sensor, regular EF mount, higher rolling shutter etc. Panasonic released a FF 6K, but without RAW and only 4:2:0. What happens when all of them start to release cameras with the current lacking parts improved, but at a much lower price point? When they all match up to Red's resolution, all have a widely adopted easy compressed RAW workflow, all have a higher readout sensor with 14-15 stops and better color science with nicer roll-offs? Coupled with things like RF-mounts, internal wireless video and so on...

The last ten years have been all Arri and Red, with a bit of Sony if we look at the most commonly used systems. But when technology catches up and the high performing tech isn't being a special sauce anymore, only regular sauce, how will companies compete? This is what I meant with Red needing to disrupt their own lineup, to match lower price-points of other brands in order to be ahead in the game. The whole mini-mag ordeal that went viral did show us that because tech is getting lower in price, premium price-points gets more and more irrational as a business model and R&D costs can't be a reason for higher price-points if others deliver the same or better performance. It was both fine and understandable that when being on top with camera technology, Red could charge premium prices, even if some tech wasn't that expensive off-shelf. But at this time, there has to be a change and I seriously hope that Komodo is that change. Because DSMC3 will have a hard time competing in the market ahead. There will have to be some major improvements that's innovative enough to compete with what's coming and I'm not sure Red will keep up with that. DSMC3 will at least need to have a premium model with a 65mm sensor of above 8K resolution at 18 stops dual-iso, all the wireless gizmos available, internal to the body. It will need to be built upon being the most reliable Red ever, dual card slots, improvement in cooling and so on. It would need to be a new chapter of quality for Red, but also the industry.

Because, if the competition, two years from now, release cameras with 8K full-frame, fast readout 14-15 stop sensors with superb color science and at least 60-90 fps max, but at a sub $10000 price-point, there will be a hard time justifying Red anywhere else but the very highest level productions (Disney/Marvel/Netflix/Amazon). Komodo could, however, be a disruptor to gain a market share before that, if done correctly.

How else can Red stay on top other than with brand recognition?

This is all you'll ever capture—a linear signal response to incoming light. This is the data, this is what's captured. There is no roll off here.

So, what is the reason for having a highlight roll-off OLPF and a low-light OLPF if everything is only software manipulation of bayer data?

 

[jamesbridges]

Would be nice.. but as of today.. CFEXPRESS is still a unicorn.

https://youtu.be/3x9W2x8bweM

https://www.bhphotovideo.com/bnh/co...earch=yes&sts=ma&Top+Nav-Search&Ntt=CFexpress

Good article:
https://www.newsshooter.com/2019/09/04/sandisk-reveals-new-speedy-type-b-cfexpress-media/

 

[AndreeMarkefors]

So, what is the reason for having a highlight roll-off OLPF and a low-light OLPF if everything is only software manipulation of bayer data?

No reason at all.

Are you referring to Skintone/Highlight and Lowlight OLPFs from RED? I don't remember there ever being a rolloff component to them. Aren't they simply different levels of light filtering, where the highlight filter cuts a bit more (problematic) light, in order to get cleaner, truer images, but at the cost of some mild light loss?

Isn't one of the reasons that RED quotes 16.5+ stops that they barely filter anything at all (or used to)? That they let it all in, including "problematic' wavelengths to increase signal strength/sensitivity?

BMD has something similar where they have a 'highlight recovery' checkbox that is off by default. You can check it and you'll recover actual data, but (strange) color shifts may occur.

Where do you suggest the 'rolloff' component lives in the signal chain? How could such a thing exist in a digital sensor?