Possibility to use 5K Mysterium-X Epic with 6K Dragon Epic for 3D?

[Asif Limbada]

I am thinking not to upgrade my EPIC-M to 6K, I might just get a EPIC-X 6K when it comes out and keep my EPIC-M 5K for time being.

The reasons are simple, FOV on 6K sensor if using 5K res, would be more tele & editing bottlenecks to work with 6K footage would be a pain. Lets hope the computers keep up with 6K footage, so should the software (Its was a pain working on CS5.5 for the past 8 months...to buggy. CS6 = Awesome).

Current RED Rockets do not handle 5K multi track timeline. One 5K track per RED ROCKET (that costs USD4000+). Lets not get me started with 5K HDRX multi track timeline. Also we are whining about Mac Pro and its connections, I think RED ROCKET is PCE 1.1 or something (correct me if I am wrong)

But I would get a 6K EPIC nevertheless. I love to have that Dynamic range and resolution. And would love to do some 3D. But when I think about it, in my head it feels impossible to do 3D with 5K MX Epic and 6K Dragon EPIC at lets say both set at 5K. Will it be compaitable.

Guys I am just thinking out loud. Would love some input.

 

[Matt Ryan]

No it won't. Even though resolution will match, Field Of View will be slightly different due to the new sensor. On top of that the sensors are different therefore they will not match image wise for 3D. Dragon is supposed to have more DR which is impossible to add to the MX. Also, there are more pixels in the Dragon sensor which will give a sharper image vs. the current MX. So it would be tough to do 3D with a mixed setup.

 

[Asif Limbada]

Ok then, why isn't RED making the camera capture the whole sensor and down converting it to lower res rather than allowing the sensor to be sensitive to light on a smaller area (or zoom) each time we go lower in resolution. Wouldn't the whole idea of using S35 or larger sensor seems pointless.

if you are shooting 4K on a 6K sensor, it does become a smaller sensor (figuratively speaking), noise floor becomes more evident (especially using 800ISO on 2KWS on the 5K Epic).

From what I understand the higher the resolution, the more bandwidth is required to transfer data from the sensor to the SSD. But if we let the whole sensor to capture light and only allow it to send 3K or 4K data through the bus, wouldn't it more practical.

I know I am not a technical guy when it come doing engineering of a camera and all the science of it, I just want to understand the logic to why RED is doing this and why can't it be possible to down convert the whole data???

 

[Asif Limbada]

I need to know why can't we get full FOV at any resolution, why does it "zoom" in. I spoke to DOP recently, who had the same query when we were renting the EPIC. He knew about it, but was very interested to know why do not we get full FOV at any res.

The only thing I can think of is that its something to do with fps and bandwidth of each res.

Anyone????

 

[Gunleik Groven]

As to your 1st question.

You can get the same FOV on both cameras, but they shouldn't match dynamically and probably not totally colorwise either, as the sensors are different.

But S35 downsampled to 2k/1080 should be the same FOV.

Also you are confusing things a bit (or maybe I am...
But the pixel density is supposed to be higher on the Dragon, which means that 6k FOV could well be = 5k FOV

Downsampled to 2k the resolution difference is probably insignificant.
I'd guess you'd be hard pushed to see a difference of significance on resolution even at 4k.
BUT the sensor lightresponse will be different, and that would make them quite bad as a matched pair for 3D

Just me speculating and interpolating here.

 

[Doug Beatty]

I would hazard a guess and say that cropping is much much less processing intensive vs. scaling in camera when you are dealing with raw images.

 

[Gunleik Groven]

Scaling is not like RAW....

When you record RAW you record the pixels available for that FOV.

Scaling is done in the next step.

Many reasons for cropping. Highspeed is one of them...

 

[Asif Limbada]

Thanks Gunleik for joining in.

But I still am confused. Why can't we have the option to down sample the footage with FF5K to 2K or 3K or 4K with max 96fps without losing the FOV.

I think that should be my final question. Is it because of the bandwidth bottleneck from the data of the sensor or the downsampling is processor hungry etc.

This is just eating my head :emote_headwall:

 

[Gunleik Groven]

Simply because it is RAW...
It wouldn't be, if one did what you sugest in-camera.

Easilly done in post, though.

Cheers!

G

 

[Asif Limbada]

Simply because it is RAW...
It wouldn't be, if one did what you sugest in-camera.

Easilly done in post, though.

Cheers!

G

Yeah, but some people just don't want 5K (or 6K when it will come) when they really want is 4K native. Why compensate on FOV

There are many different ways to do downsample in post, I do it all the time. But I want to know the reason why RED is using cropping rather than downsampling.

 

[Asif Limbada]

I spoke to another DOP, same question on why we can't we down sample the FF5K to 4K on the camera.

 

[Timothy Carr]

Am i the only one thinking that if it scaled down, we would get AA issues?

 

[Björn Benckert]

Epic and scarlet "only" has one downscale and that is to SDI/HDMI HD...
To do advanced downscale in camera is very costly / CPU demanding and especially if you want to do it at high frame rates.
Alexa and other cameras do this, and for that reason among others the price tag is higher.

R3D is RAW from the sensor and has it's benefits. while a "scaled raw" has other benefits, like higher frame rates with less compression per pixel etc.

Using two cameras with different sensors for 3D is actually quite simple and your eyes are quite adaptive to dynamic range differences and such. the actual sensor size is off course something that would need to be dealt with in post but it's only a constant scale/crop thats needed for one of the feeds. So that easily taken care of. However with all the hassle of shooting 3D it might be one of those things you do not want to add to the injury.

Red can quite easily setup corresponding frame rates and pixels usage so that it's possible to shoot 3D with even one scarlet and a dragon at 4k or lower resolutions. Downgrading imagery is quite easy so it would get the look of shooting 2x scarlets and not 2x dragon.

Personally I hope the wave of stereoscopic / 3D will pass I can not for the life of me think of any one move that looks better, or leaves a better impression in 3D than mono. For me it's just a wow effect the first minute and then it bugs my eyes for the rest of the move... but that might be just me.

 

[Timothy Carr]

For me it's just a wow effect the first minute and then it bugs my eyes for the rest of the move... but that might be just me.

*Movie

 

[Asif Limbada]

Epic and scarlet "only" has one downscale and that is to SDI/HDMI HD...
To do advanced downscale in camera is very costly / CPU demanding and especially if you want to do it at high frame rates.
Alexa and other cameras do this, and for that reason among others the price tag is higher.

R3D is RAW from the sensor and has it's benefits. while a "scaled raw" has other benefits, like higher frame rates with less compression per pixel etc.

Using two cameras with different sensors for 3D is actually quite simple and your eyes are quite adaptive to dynamic range differences and such. the actual sensor size is off course something that would need to be dealt with in post but it's only a constant scale/crop thats needed for one of the feeds. So that easily taken care of. However with all the hassle of shooting 3D it might be one of those things you do not want to add to the injury.

Red can quite easily setup corresponding frame rates and pixels usage so that it's possible to shoot 3D with even one scarlet and a dragon at 4k or lower resolutions. Downgrading imagery is quite easy so it would get the look of shooting 2x scarlets and not 2x dragon.

Personally I hope the wave of stereoscopic / 3D will pass I can not for the life of me think of any one move that looks better, or leaves a better impression in 3D than mono. For me it's just a wow effect the first minute and then it bugs my eyes for the rest of the move... but that might be just me.

So Bjorn, in terms of CPU power, the FF5K FOV cannot be down sampled to 4K on camera due to CPU power required to do it. But C300 not only takes in data from S35 sensor at 4K and down samples it to 2K or 1080p (I dont care what and how it acquires) and compresses that particular data into 8bit to the memory.

Now, EPIC is a superior camera and why can't it do it.

I think there should be 2 options for lower res options, like scaled or downsampled. Scaled footage gives higher fps and downsampled gives a maximum of 96fps.

 

[Timothy Carr]

So Bjorn, in terms of CPU power, the FF5K FOV cannot be down sampled to 4K on camera due to CPU power required to do it. But C300 not only takes in data from S35 sensor at 4K and down samples it to 2K or 1080p (I dont care what and how it acquires) and compresses that particular data into 8bit to the memory.

Now, EPIC is a superior camera and why can't it do it.

I think there should be 2 options for lower res options, like scaled or downsampled. Scaled footage gives higher fps and downsampled gives a maximum of 96fps.

Its just the way red deign there cameras. If you dont like it, use something else...

 

[Björn Benckert]

So Bjorn, in terms of CPU power, the FF5K FOV cannot be down sampled to 4K on camera due to CPU power required to do it. But C300 not only takes in data from S35 sensor at 4K and down samples it to 2K or 1080p (I dont care what and how it acquires) and compresses that particular data into 8bit to the memory.

Now, EPIC is a superior camera and why can't it do it.

I think there should be 2 options for lower res options, like scaled or downsampled. Scaled footage gives higher fps and downsampled gives a maximum of 96fps.

The options are out there, I think the lowest priced one is the ninja, it records the 1080p stream from your SDI out.... but only up to 30fps...

Even though I understand what you mean, I think, or see, that it's not top priority at RED, and a C300 only does what it does on certain frame rates. It's not so likely that the canon camera will grab a 5k frame at 400fps take all that data and downsample it in any good way to 1080p. And if it does it will not preserve the raw colors.

But I have no doubt that RED in a not so distant future will make a down sample module, that can pass down low resolution images to a second card with higher or lower compressions.

Using the full sensor at all frame rates would off course be nice, but I find it good that red gives us what they can when they can. Now we got a big sensor with a lot of pixels in a small package. Then they will add all the other features when they have them ready. And remember all the other cameras out there still is leaps behind, too big and or to expansive. I think Red is on the right path with their modules and making their cameras as different brains using the same modules. I think they guys buying the C300 are not looking forward for the day when the C350 or whatever it will be called is sold. With the epic, we can look forward to get the dragon and probably a few other upgrades down the line.

 

[Asif Limbada]

Its just the way red deign there cameras. If you dont like it, use something else...

Tim, where in the entire post did I say I did not like it. It sounded to me as rude and very defensive, please, this is a healthy discussion, lets keep it this way thanks.

Also, my main aim is not to take a jab at RED. My question is a simple, why not "down sample". There should be a logical reason to that, Bjorn, helped a bit with his explanation.

 

[Timothy Carr]

Tony, where in the entire post did I say I did not like it. It sounded to me as rude and very defensive, please, this is a healthy discussion, lets keep it this way thanks.

Also, my main aim is not to take a jab at RED. My question is a simple, why not "down sample". There should be a logical reason to that, Bjorn, helped a bit with his explanation.

Its Tim, and sorry if it came across all harsh, i didnt mean it like that. I was just saying, there is a reason there like that. Down scaling at reasonable quality, at good FPS while maintaining all the extra goodies like HDRx, etc, in realtime, while writing to a SSD takes alot of power. BUT its a hot topic, and im sure red will give us a nice way of doing this in the future

 

[Asif Limbada]

The options are out there, I think the lowest priced one is the ninja, it records the 1080p stream from your SDI out.... but only up to 30fps...

Even though I understand what you mean, I think, or see, that it's not top priority at RED, and a C300 only does what it does on certain frame rates. It's not so likely that the canon camera will grab a 5k frame at 400fps take all that data and downsample it in any good way to 1080p. And if it does it will not preserve the raw colors.

But I have no doubt that RED in a not so distant future will make a down sample module, that can pass down low resolution images to a second card with higher or lower compressions.

Using the full sensor at all frame rates would off course be nice, but I find it good that red gives us what they can when they can. Now we got a big sensor with a lot of pixels in a small package. Then they will add all the other features when they have them ready. And remember all the other cameras out there still is leaps behind, too big and or to expansive. I think Red is on the right path with their modules and making their cameras as different brains using the same modules. I think they guys buying the C300 are not looking forward for the day when the C350 or whatever it will be called is sold. With the epic, we can look forward to get the dragon and probably a few other upgrades down the line.

No Bjorn, I need FF5K FOV with 4K native r3d at 96fps. When I want 120fps I just will have the option to select "Scaled" 4K.

Having worked with RED RAW, I would not use ProRes anywhere in my workflow