Welcome to our community

Be a part of something great, join today!

Register Log in
  • 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 :)

Epic, dragon, efp/eng.. please adress the need for a more sensitive 5k sensor !

D

Denis Buhot

Well-known member
Joined
May 1, 2007
Messages
381
Reaction score
0
Points
0
There has been some comparison tests out by now. Dragon vs MX. Thanks to all who conducted the tests for collective benefit. However great the improvement is over MX, - it really is – it also has to be said that dragon is definitely not the ultimate sensor for efp/eng/wildlife users, simply because iso improvement/noise control is in no way the best part of what it delivers. Rather marginal improvement here, as compared to much extended DR and color rendition. And that is not undermining Dragon at all. Anyone that denies that will end up simply demonstrating he is not likely to shoot in very scarce, natural light. Dragon is perfect for stage use and controlled lighting environment. A priority at Red that has always been apparent.

To me, in line with my personal needs as a red user, the question seems to be : will red leave the efp/eng market to arri and possibly others ? In spite of epic remarkable compact format, modularity and light weight, a combo that makes it so wonderful to use out of stage in every aspect but sensitivity ? Will red ALSO adress the need for a 4/5 k (maximum) sensor with definitely higher low-light capability, in addition to delivering new sensors with higher resolution than 6K ? I have put my trust in red since the beginning, I have seen and benefited from its past achievements, and I really hope it will !
 
As an extensive mobile EFP style user of Epic, and multiple other camera systems, I largely agree with you Dennis. For natural light shooting of sports, wildlife, documentaries, adventure travel, cultures, and a myriad of other mobile natural light genres, good low light shooting sensitivity is absolutely critical. Unless the delivery for these genres is IMAX ( a rare delivery), then acquiring the field footage in 4k or UHD is just fine on resolution, but low light sensitivity is critical. Dragon is a great sensor in so many ways - but not necessarily great for mobile EFP field production in frequent low light environments. That said, the upgrade to Dragon for almost all Epic users is to me a "no brainer", unless almost all their production is mobile, in low natural light.

So as not to re-post, here are some recent posts of mine on other Red User threads that address some of the same concerns you have:

"ARRI Amira pricing" (see my posts number 44, 52, and 73)
http://www.reduser.net/forum/showthread.php?111984-Arri-Amira-Pricing

"Red Focus" (see my post number 209)
http://www.reduser.net/forum/showthread.php?111859-RED-FOCUS/page21

You'll see in my post on the "Red Focus" thread that I documented that in the very beginning of Red there was almost an equal intention to support both the cine and mobile EFP industries. But later on more and more of the EFP features and capabilities were dropped from Red's emphasis. Then the Fixed Lens Scarlet disappeared, which to me signaled a significant shift by Red away from EFP support in favor of cine support.

My own personal opinion on this as someone who was there giving EFP input to Red from the very ground floor in mid-2005 to the present day, is that Red has significantly evolved from an original intent to support both cine and mobile EFP, to a present focus of almost entirely cine support, with mobile EFP being the afterthought - in essence the "Red-headed stepchild" (pun intended!)

That said, Red has to do what they feel is best for the company, and I get that. But it is frustrating to mobile EFP users who have supported Red all along from the beginning. Jim, Jarred, and Red Team are all good friends and I respect them greatly. That said, I've always told them what I think they need to hear, which hasn't always been what they wanted to hear, or followed.
 
Steve, I had missed those posts...

Steve, I had missed those posts...

Steve Gibby said:
My own personal opinion on this as someone who was there giving EFP input to Red from the very ground floor in mid-2005 to the present day, is that Red has significantly evolved from an original intent to support both cine and mobile EFP, to a present focus of almost entirely cine support, with mobile EFP being the afterthought - in essence the "Red-headed stepchild" (pun intended!)

That said, Red has to do what they feel is best for the company, and I get that. But it is frustrating to mobile EFP users who have supported Red all along from the beginning. Jim, Jarred, and Red Team are all good friends and I respect them greatly. That said, I've always told them what I think they need to hear, which hasn't always been what they wanted to hear, or followed.
Click to expand...

You made your point ! A rather sad point, to me, too. It is all a matter of strategy, I hope... not despise for efp/eng as a minor field. To me, that will definitely contribute in making Red an "american movie camera", and let Arri dominate the market again in its european strongholds...
 
But you have to ask yourself about the cameras that reportedly are a lot more sensitive -- i.e. the Canon's -- whether some noise reduction is happening in their recording chain. I don't doubt that they are actually more sensitive, but it may only be by an actual stop or two at the most, the rest is mostly noise reduction to allow speeds above 6400 ISO that look acceptable in noise level.

It would be a good test, take a 6K Dragon frame shot at 6400 ISO and apply noise reduction and downscaling to see how it compares to a 1080P image from a Canon C3000 shot at 6400 ISO. An ENG shooter may find that if he is only delivering 2K/1080P then noise-reduced 6K Dragon footage shot at high ISO's would work for his low-light needs. I think the resolution hit from heavy noise reduction would be fine in this case, the question really becomes about shadow detail, whether it would fall off too much compared to a Canon.
 
yes...

yes...

David Mullen ASC said:
But you have to ask yourself about the cameras that reportedly are a lot more sensitive -- i.e. the Canon's -- whether some noise reduction is happening in their recording chain. I don't doubt that they are actually more sensitive, but it may only be by an actual stop or two at the most, the rest is mostly noise reduction to allow speeds above 6400 ISO that look acceptable in noise level.

It would be a good test, take a 6K Dragon frame shot at 6400 ISO and apply noise reduction and downscaling to see how it compares to a 1080P image from a Canon C3000 shot at 6400 ISO. An ENG shooter may find that if he is only delivering 2K/1080P then noise-reduced 6K Dragon footage shot at high ISO's would work for his low-light needs. I think the resolution hit from heavy noise reduction would be fine in this case, the question really becomes about shadow detail, whether it would fall off too much compared to a Canon.
Click to expand...

there must be some truth in that, indeed. May be canon C500 or sony F55 would be more appropriate, anyway, I don't know... Worth trying.

Anyway, to me, the silence from Red people is a matter of concern.
 
Good points about noise reduction Dave - I agree with you. I've also felt that in-camera noise reduction was a non-publicized factor in the low light performance of many of the new cameras. It would be interesting to test that out further...

When me and my crews use my Epic cameras for our usual mobile EFP style of field production, with the intended delivery being ATSC 1080 television, we generally shoot in Quad HD (UHD) resolution to help make the footage evergreen looking forward, for the good mathematical scaling and resampling of UHD down to 1080, and to master out in UHD for the coming move to UHD TV. As such, acquiring the field footage in 6k, or even 5k, is overkill for our delivery medium - now in 1080, or future in UHD. Thus in mobile EFP style production genres acquiring footage in any resolution above 4k is essentially beyond the point of diminishing returns. But the need for very good sensitivity in the frequent low natural light conditions we shoot in drives us to seek sensitivity over ever increasing resolutions.

I think what Dennis is getting at, and I agree with him, is that the sensor performance needs of Mobile EFP users of Red cameras are often quite different than cine style users. We love high resolution up to a point (usually 4k or UHD), but the optimum sensor for the mobile EFP industry, especially looking at our current move to UHD mastering and pending UHD TV delivery, is for a 4k/UHD resolution sensor which is optimized for great natural low light sensitivity - and frankly as good as the Dragon sensor is, it is targeted at cine applications over mobile EFP applications. Thus the frustration here among longtime adopters and supporters of Red whose work is primarily mobile EFP. That's driving us out to seek, test, and use other emerging camera systems optimized for mobile EFP work - such as F55, Amira, etc.

Hopefully Red is listening closely to this input from us mobile EFP users. In terms of sheer number of productions per year, number of people employed, and dollar amount of camera equipment bought each year the mobile EFP industry is the largest portion of the motion media industry. IMO it makes sense for Red to listen closely to their mobile EFP users, and then design a separate sub-line of EFP optimized sensors. The new EFP targeted Red cameras using those sensors would then have internal ND, a broad range of traditional EFP camera busses and and features, and Red should offer a removable lens version to compete with the F55, and a lower priced fixed lens version, a unibody one with a great 20x optical zoom, as a B camera for the removable lens camera. Convertible servo zoom capable S35 zooms like the Fujinon Cabrios I've been testing out are perfectly suited for a mobile EFP production with a Red 4k/UHD removable lens camera like I described. Then if we want to shoot cine style its as easy as quickly taking the servo zoom unit off of the lens, adding in a MB, FB, and FIZ, and we're all set.

Whether the above will happen is up to Red, and they need to do what they feel is best for the company. It will be interesting to see how this all plays out...
 
Apologies if the following is inept - it's not an area I've delved into ...


If the RED sensor is being reset, and exposed for half of the frame, and then read out - maybe there is something that can be done to increase the signal-to-noise for those low light scenarios?


+) Record at twice the frequency.


+) Main track at 355 degrees (almost exactly the same physical time as a regular recording)


+) Record X track at 5 degrees ( ie 1/71 of the the frame)


Where's before there is only frame to process (F_old) we now have 4:
F_A = long frame ~= F_old
F_a = mini frame that can be used to dynamically calc fixed pattern noise
F_B = long frame = useful light data thrown away (up to 6dB more sensitivity)
F_b = mini frame that can be used to dynamically calc fixed pattern noise


The noise reduction could start with:
F_a_FPN = adaptive_alg_for_this_pixel_position( F_A, F_a)
F_b_FPN = adaptive_alg_for_this_pixel_position( F_B, F_b)


F_A_improved_sig2noise = F_A - F_a_FPN
F_B_improved_sig2noise = F_B - F_b_FPN


For mostly statics shots average for an additional 6dB noise reduction.


For moving images clever software with 3d motion tracking required to not smear.


As resolution is being ~halved, smearing can probably be reduced by a third (maybe by up to 50%).


Make sense?


AJ
 
Who needs more? Where are the boundaries? I think we are allready in the "bonus land". The tool we have in our hands don't hinder from making strong pictures any more. Use your creativity, get around limitations. ISO are numbers, do we really need more? I think we could still emproove ergonomics, not quite sure about the need of more iso.

Pat

 
There are some known ways to increase the sensitivity -- they could keep the same size as the 6K sensor but only make it 3K with photosites that are twice the size, for a 2K/1080P finish. They could sacrifice a little color reproduction by using a few clear photosites in the CFA such as some companies are exploring for smaller sensor cameras. But all of this would probably go against Red's interest in making cinema-quality high-resolution images.

Sequential exposure read-outs is a bit problematic in terms of motion.
 
Red said long ago that they felt 1080 wasn't in the future for them, but they've been a champion of 4k (and Quad HD/UHD) from the beginning in 2005, with their original plans for additional 1080 canceled in 2007. If they took the 6k Dragon sensor, kept it the same size, but made it 4k with appropriate adjustment in photosites, designed for a future-proofing UHD finish, and increased sensitivity, that would be exactly the parameters which would seriously satisfy mobile EFP users of Red cameras. That would also fit into Red's stated interest in calling 4k as the threshold resolution for the present and future. Red camera users doing Cine and IMAX delivery work could opt for the Dragon 6K sensor optimized for that - and mobile EFP Red camera users could opt for the 6k/4k optimized sensor to fill their needs. Best of both worlds, no significant R&D needed, both industries satisfied, and greatly increased camera sales for Red - so everyone wins!
 
Let me preface this by saying what I am about to describe is a wishlist, not a demand on Red as I do believe they are still young (relative to a company like Arri) and have their hands full as it is.

These are some of the things I think might be cool for a RED EFP camera that could be very well suited for mobile crews capturing images for documentary, corporate, etc.
  • UHD resolution of 3840x2160
  • An ergonomic body design that fits your shoulder like a glove
  • An advanced EVF that LUTs can be applied to with great color accuracy. Also, I really like the Amira sliding handle/EVF mount. A person operating the camera should very confident in the image they are seeing to judge color, and tonal values. It should almost be assumed there will be no external monitor to judge the image quality.
  • I would love to see Red collaborate with Sound Devices and integrate an in-camera system with quality mic pres, low noise, and efficient power. External pots a must. For those times when there is a sound person on the crew, no problem they can feed line level and feel confident that the sound quality will be top notch. If they are recording to an external device it can be a backup. But for those run and gun times, the camera will have a high quality sound recording system.
  • Advanced cooling system that is virtually silent. Camera noise should be a non issue for quiet, emotional interviews.
  • 120fps at full quality. I still believe that a specialized camera such as Phantom is a better choice when you get up into the really high frame rates. Obviously, if they can pack more FPS and keep the high quality, so be it.
  • REDCODE RAW and REDCODE RGB modes. For those who want to play a lot with the image in post shoot RAW. For those who don't, shoot REDCODE RGB. I know it's not in RED's DNA to offer this, but perhaps a 1080P RGB mode would be nice to save on media storage. But if not, at least a UHD RGB mode.
  • Physical buttons on the camera for all of the most needed functions. Menus for all those things you don't change that often.
  • Internal NDs up to .9. with built in IR tuning. Perhaps with a way to go completely clear and add your own NDs and IR solution.
  • Extremely accurate metering. For example, there could be a false color mode specifically for the REDCODE RGB that would allow you to deliver images with confidence for those times when you literally walk away from the job at the end of the day and someone else will be handling the image from that point.
  • Built in noise reduction. This one might fly in the face of RED. But rather than forcing it on people you have a way of turning it off in camera. So you have the choice to do it as you shoot the scene and determine if it helps you push the image a little bit or save it for post. But this could help with those low light situations.
  • Next generation sensor tech. Based on the UHD sensor size, get the most light sensitivity you can so the camera performs really well in typical EFP situations. Along with the NR, it could then be possible to achieve ISO 6400 relatively clean and I think this will be very attractive to people.
  • Flexible lens mounting options. Canon, Nikon, PL, etc. Not sure if RED is in the lens business anymore, but perhaps they could team up with one of the heavy hitters to offer a lens built specifically to take advantage of this camera system. Something in the 16-85mm range with aperture control, a zoom rocker, etc and around an F2 or F2.5. In other words, a Cabrio that is perhaps even more integrated with the camera, perhaps storing metadata in the file.
  • Power options. Provide v-mount and AB plates that are built into the ergonomics of the camera. Also provide power for accessories.
  • Standard I/O. Full size BNC. Full size XLR (if space is an issue, I'm okay with TA3 connectors).
  • Timecode. This would be a great opportunity to try and partner with some of the companies that specialize in this. The Alexa is pretty good, but I think timecode could be even better on all of these cameras. Timecode jamming should be easy, dependable, and a non-issue.
  • Advanced focus tools. We are starting to see new approaches to this but as in a lot of technological advances, the early tools tend to be bulky. At the very least, have some really great peaking functionality. When shooting on the fly with large sensors, camera ops need tools that allow the to asses an image quickly and make almost immediate decisions. So something like focus tools should be built in a way that gives camera op immediate feedback to what is in focus and with the advanced EVF mentioned above should be a dream scenario for an op.
Obviously, some of this is pie in the sky thinking. And you'll notice I mention "advanced" a lot. What I mean by that is for the engineers, designers, and principles at RED to have a look at these things, identify (with the help of a lot of real world feedback) things that don't work well, align themselves with companies that can offer solutions that do work really well (or if they can develop something equally great internally that is fine too), and create advanced tools that could really aid camera operators who don't always have a large crew with them such as multiple camera assistants, sound recordists, etc. When I say advanced cooling, I truly mean it has to keep the camera at regulated temperatures and make no noise (at least no noise that a microphone can pickup). These are not easy solutions, especially if you want to keep the cameras in line with today's pricing.

Again, RED has done an amazing job. This is a wishlist based on an EFP style camera. Whether this is something that makes sense for them to do in the near future, I don't know. And I'm sure not everyone will even agree with the things I just mentioned. Everyone's vision of the ideal mobile camera may be different. And I'm not a camera designer, so....
 
Very well thought out list Steve...most of those are essentially the rough features I'd recommend for such a camera, and many of them I've been recommending to Red directly since 2005, and on Red User since 2006.

To save on R&D and tooling the form factor could be kept modular like the regular Epics, but have new specific modules to enable the form factor, buses, and utility of mobile EFP cameras.
 
David Mullen ASC said:
Of course, if the recorded area was 3840 x 2160 raw, then you'd get the arguments that this was not "true 4K" in terms of measurable resolution, though it would make an excellent 2K/1080P RGB downsample.
Click to expand...
In this case I might consider it as a perfect oversampling for 1080p with a nice UHD for average screens of 60". And it could be very respectable on big screen. If the IQ is really good I think it would translate well. Alexa 2.8K looks good to me most of the time so the potential is there.
 
David Mullen ASC said:
Of course, if the recorded area was 3840 x 2160 raw, then you'd get the arguments that this was not "true 4K" in terms of measurable resolution, though it would make an excellent 2K/1080P RGB downsample.
Click to expand...

That is true.

And I suspect there are other ways than bigger photosites to calibrate a camera for lower light. Problem is that a sensor has a fixed DR, and thus you would have to give up highlight stops to get lowlight stops.
In the "cinema" world I know, that would be a hard sell...

But I can see the use for it in other contexts.

Making "the same sensor with bigger pixles", is as far as I can understand things I know little about, much more complex than any other suggestions, as it is essentially a new sensor-program. Much harder (I would think) than modifying the body.

I would guess a re-calibrated "same" sensor would be more doable.

But gee... I am speculating now. :)
 
All good points...then how about a re-calibrated, 5k sensor based on the Dragon 6k sensor, that shoots in 5kHD resolution - 16x9 aspect ratio, which when debayered makes for excellent UHD finish, and naturally a great resample for 1080 finish. All this engineered down from the existing Dragon 6k sensor.

Two sensors: the existing Dragon 6k sensor for cine/imax applications, and the re-engineered Dragon 5k sensor for UHD, and 1080 applications. Then in the words of Phil Robertson "Everybody happy, happy, happy!"Good
 
I have a better suggestion, how bout a UHD RGB recording module?
It would use the same Red mini-mags, but record a LOG or Rec709 Redcode RGB in UHD res but it would be derived from the full 6k or 5.5K raster, but the real advantage (besides downscaling for resolution) would be onboard noise reduction. I feel it's very important not to have onboard NR in the RAW imaging plane for cinema use where you want every bit of detail possible and post NR is de rigueur.
However for ENG/EFP and episodic TV having a clean high ISO edit ready UHD file would be extremely useful and is totally possible with the current image stream coming off the Dragon sensor.
 
Last edited:
Sounds like a workable, and relatively easy to achieve solution Evin...

Hopefully Red is listening, because today on this thread, and the past few days on a few other threads in the Off Topic forum there has been some very good input given about the needs of the mobile EFP community
 
Steve Gibby said:
...how about a re-calibrated, 5k sensor based on the Dragon 6k sensor...
Click to expand...

The proposed Scarlet Dragon uses a 5k sensor based on the Dragon 6K sensor. If the sensor can be re-calibrated to maintain the current MX highlight clipping level while adding more useable stops at the lower end...why not?

Not sure how you could add internal ND's without altering the current Scarlet/Epic form factor though.

Wondering too about not being able to delete clips in-camera. Is that something people out in the field could live with or would it be a deal-breaker regardless of other changes?
 
You must log in or register to reply here.
Back
Top