Welcome to our community

Be a part of something great, join today!

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 :)

#21...

Status
Not open for further replies.
Oliver Skibbe said:
My guess: they use a kind of spatial HDR, where pixels have different sensitivity.

HDR-Pattern.jpg
Click to expand...

This together with Graeme's math makes perfect sense to me.

Is this it Jim? Come on!

Arri doesn't read REDuser... everybody knows that! :)
 
The HDR weekend, YHEAA! Team.

#21 sounds amazing.

I always thought that if apple or someone give me ISO and FLUT in secondarys room I can squeeze some stops, but HDR on camera or with RED soft would be amazing.

Some smart exposition aplication on RCX??? ;-)
 
Oliver Skibbe said:
HDR-Pattern.jpg
Click to expand...

I posted that pattern 2 years ago, posing the same question. At that time it made sense. But what we know today makes me strongly doubt that Jim would allow for a permanent decrease of resolution when HDR is not used. It's kinda like lineskipping. And we know Jims oppinion about that.

If you however make a system to read the whole sensor data twice (i dunno how. Dual sensor read circuitry perhaps). Each with different iso rating, you could achieve full resolution HDR with no drawbacks other than noise perhaps.
 
But think of the strong Graeme sauce used by RED, and about how much resolved detail there is in the footage compared to what is expected from Bayer cameras. Also, with a 5k sensor, there is a whole K to throw away and still satisfy Jim's 4k standard... The thing is, if this were the case we would find out, certain people will be taking a very close look at the sensor and they will be able to tell.
 
RED has always permitted that, it is the nature of RAW, which allows you to interpret the data however you want. You already can repeat and blend the whole thing in both 250 and 800.

However that is not HDR, which requires two or more different measurements. Same goes for what Alexa does which was suggested as HDR by some of Arri's marketing, which was what you are describing, in-camera, at a hardware level. Not HDR.

For HDR to have a reason to exist as a technical term it needs to describe something different, not the same thing with fancy post massaging applied (be it digital or hardware massaging, done in-camera or in a post suite).

HDR then requires you to get at least two measurements that differ by ways of a different native sensor's sensitivity; different shutter speed (but motion blur is affected); a different aperture (but DoF, sharpness, contrast, etc, would be affected --if-- there was a way to even do that); or different filtration (like the chart posted by Oliver).

However, the one based on two sets of different filtration of photosites also requires some crazy-serious mathematical magic to minimize resolution loss to a level that RED would consider "no-compromise".

In short, alien voodoo.
 
Nik Skjøth said:
I posted that pattern 2 years ago, posing the same question. At that time it made sense. But what we know today makes me strongly doubt that Jim would allow for a permanent decrease of resolution when HDR is not used. It's kinda like lineskipping. And we know Jims oppinion about that.
Click to expand...

If every single photosite on the sensor is switchable between low ISO (25-250) and high ISO (200-2000) then you can change the sensor pattern from from a classic Bayer mode (either low or high sensitivity) to a mixed HDR-pattern. Resolution in non-HDR mode will be unchanged. (Remember the RED-MX uses the same sensor in ISO 200-2000 mode.) There might be a small decrease of resolution, when the HDR pattern is used though.
 
I'm wondering if it is as "simple" as something akin to the dual gain architecture of the Alexa. With small lead in DR Arri have got using that, it's pretty clear that RED have the superior chip. If RED were to do their own system similar in theory to Arri's these kind of numbers seem to make sense.
 
Nik Skjøth said:
Dual readout from the bayer sensor. User selectable variable iso ratings, combined into HDR image.
Click to expand...

Dual readout would require two (layered?) photodiodes for each photosite. It would also require twice the number of output amplifiers and ADCs. Is the MX sensor is really that complex? I have no idea, maybe it is. Its autofocus capabilities (phase detection?) could be a clue.
 
ISO is not a setting, it is a measurement.

You don't "change" ISO, you only measure a sensor or film stock's ISO which means that the sensor gives its best images at that sensitivity, or you push and pull the negative through chemical processes and the digital negative through digital filters to try to make it look like it was shot with a sensor or film that had a different ISO... but, in the pushing and pulling you lose data and accuracy so the image degrades.
 
Roberto Lequeux said:
ISO is not a setting, it is a measurement.

You don't "change" ISO, you only measure a sensor or film stock's ISO which means that the sensor gives its best images at that sensitivity, or you push and pull the negative through chemical processes and the digital negative through digital filters to try to make it look like it was shot with a sensor or film that had a different ISO... but, in the pushing and pulling you lose data and accuracy so the image degrades.
Click to expand...

Yeah I suppose my mistake here was to assume, that a photosite has no limit as to how much/little light it can accept. (Since Jim talket about the NO NATIVE iso rating.) Obviously it must. (duh,:emote_headwall:)

I assumed that the limitation was in the read curcuit, hense dual readers made sense. And eventho you can change ISO in post, you would still be limited by the 13 stops of recorded information. By dual reading you would have up to 26. Oh well.
 
I suspect they're doing something like reading the sensor out multiple times for every frame, and they've got some magic for putting together the data they get in a way that doesn't make a mess out of in-frame motion.
 
Chris Kenny said:
I suspect they're doing something like reading the sensor out multiple times for every frame, and they've got some magic for putting together the data they get in a way that doesn't make a mess out of in-frame motion.
Click to expand...

I think it is related to the dual ISO range feature. Can the camera perhaps read both ranges simultaneously for a single exposure and interpolate the data to produce an HDR image?
 
OT -- but I assume everyone experienced the "Reported Attack Page" error that stopped me from checking out the forums. Finally found out on CML what was going on. One more reason for Jim (or RED) to setup a twitter feed we can follow...
 
Nikhil Kamkolkar said:
OT -- but I assume everyone experienced the "Reported Attack Page" error that stopped me from checking out the forums. Finally found out on CML what was going on. One more reason for Jim (or RED) to setup a twitter feed we can follow...
Click to expand...

I thought someone reported RED as a scam - which would be nothing new ;)
 
Nikhil Kamkolkar said:
OT -- but I assume everyone experienced the "Reported Attack Page" error that stopped me from checking out the forums. Finally found out on CML what was going on. One more reason for Jim (or RED) to setup a twitter feed we can follow...
Click to expand...
Reduser is supported by advertisers who benefit when people see their message here. May be one reason why REDs don't twit. Just guessing.
 
Status
Not open for further replies.
Back
Top