Magic Lantern Improves 5D Mark III Dynamic Range to 14 Stops

[michi]

This is nothing real new and have been discussed at DPR years ago.

This document presents a sensor scanning trick that results in significant dynamic range im- provements, at the cost of vertical image resolution and aliasing, especially in highlights and shadows. The ISO (analog amplification) is alternated for every two scanlines, between two user-defined values (say ISO 100/1600), so half of the image is exposed for highlights and the other half is exposed for shadows.

Not sure what point you are trying to make. This feature hasn't been implemented yet. I would say that alone is new and newsworthy. A document released years ago doesn't do us photographers in the field any good...

 

[RGF]

Great job ML.

But it´s a shame that Canon is not able to improve their products.

Because of course they have never improved their products. : there is no improvement ever in any product they produce.

This is a really ground breaking innovation. Does not happen often. I am not surprised that an outsider created it; they need to do something different to make a living. Canon can continue with incremental improvements and be dumb, fat, and happy :

 

[Quasimodo]

Dynamic range is Probably more Important to me than noise reduction or pixel count.
I think the way to do this is to split the sensor into highlights and shadows...
I rather have an 8mp 16-stop split sensor, than a 16mp one with less DR.
It can be done.
It is interesting that most HDR cameras take 3 exposures but nobody thought about changing the ISO Instead.
I believe in pushing things to the limit and opening up the possibilities of a device...
Like my jailbroken iPhone 5. Which is amazingly more useful and fun than the limited iOS from apple.


Kudos to magic lantern!
Keep up the great work!

Just curious since I have a iPhone 5. How to jailbreak it, and what is the added benefits?

 

[poias]

Dynamic range is Probably more Important to me than noise reduction or pixel count.
I think the way to do this is to split the sensor into highlights and shadows...
I rather have an 8mp 16-stop split sensor, than a 16mp one with less DR.
It can be done.
It is interesting that most HDR cameras take 3 exposures but nobody thought about changing the ISO Instead.
I believe in pushing things to the limit and opening up the possibilities of a device...
Like my jailbroken iPhone 5. Which is amazingly more useful and fun than the limited iOS from apple.


Kudos to magic lantern!
Keep up the great work!

why not 36Mp and the best DR ?

D800/E is already 36Mp with 14.5 stop DR... without the ISO tricks

 

[Famateur]

This is nothing real new and have been discussed at DPR years ago.

This document presents a sensor scanning trick that results in significant dynamic range im- provements, at the cost of vertical image resolution and aliasing, especially in highlights and shadows. The ISO (analog amplification) is alternated for every two scanlines, between two user-defined values (say ISO 100/1600), so half of the image is exposed for highlights and the other half is exposed for shadows.

Not sure what point you are trying to make. This feature hasn't been implemented yet. I would say that alone is new and newsworthy. A document released years ago doesn't do us photographers in the field any good...

LOL...yeah. We've known about cold fusion on paper for a long time. When some physicists finally pull it off?

"Meh...we've known about that for decades." :

A project like Magic Lantern is a great example of enthusiasts making a good thing even better -- for the sake of making it better. I applaud the work they're doing. They're like hot-rodders trying to squeeze every ounce of torque and horsepower out of the engine they've got...and then letting everyone drive it.

 

[AAPhotog]

Dynamic range is Probably more Important to me than noise reduction or pixel count.
I think the way to do this is to split the sensor into highlights and shadows...
I rather have an 8mp 16-stop split sensor, than a 16mp one with less DR.
It can be done.
It is interesting that most HDR cameras take 3 exposures but nobody thought about changing the ISO Instead.
I believe in pushing things to the limit and opening up the possibilities of a device...
Like my jailbroken iPhone 5. Which is amazingly more useful and fun than the limited iOS from apple.


Kudos to magic lantern!
Keep up the great work!

why not 36Mp and the best DR ?

D800/E is already 36Mp with 14.5 stop DR... without the ISO tricks

But I don't own a D800, so I couldn't care less.
What does the oh so precious d800 have to do with this topic again? Maybe 14.5 stops and 46megapixels aren't the ONLY things we'd like in a camera.

 

[jrista]

If you really want to have 14 stops of DR... just buy a D800.

Not with those tiny pixels, thank you. I edited 1500 images from one the other day and I wasn't impressed.

why not, d800 has the same FWC as 1dx but 36 Mp
The best SLR sensor on the market today

I'm sorry, but that is a bold faced lie! The D800 has a full well capacity of 44k, while the 1D X has a full well capacity of 90k. That is a TWO FOLD difference between the two, bub!

 

[jrista]

Dynamic range is Probably more Important to me than noise reduction or pixel count.
I think the way to do this is to split the sensor into highlights and shadows...
I rather have an 8mp 16-stop split sensor, than a 16mp one with less DR.
It can be done.
It is interesting that most HDR cameras take 3 exposures but nobody thought about changing the ISO Instead.
I believe in pushing things to the limit and opening up the possibilities of a device...
Like my jailbroken iPhone 5. Which is amazingly more useful and fun than the limited iOS from apple.


Kudos to magic lantern!
Keep up the great work!

why not 36Mp and the best DR ?

D800/E is already 36Mp with 14.5 stop DR... without the ISO tricks

Again, more misinformation. The D800 has 13.2 stops of native dynamic range (the dynamic range at it's full resolution). It is only able to achieve 14.3 stops (not 14.5) when downscaling from 36.3mp to 8mp (an overall LOSS in detail of over 200%!!!)

This is why DXO's reports are so misleading. From a NATIVE CAPABILITY standpoint, the D800 is a 13.2 stop camera. Depending on how much you downscale, you might gain DR via a reduction in noise, at the cost of original detail. In other words, it is impossible to get 14.3 stops of DR at the native resolution of 7360 x 4912 pixels. Given that when editing a RAW photo, you ALWAYS edit at full resolution (i.e. the extreme shadow pushing we see in something like Lightroom), it is only valid to say that the D800 is a 13 stop camera, not a 14 stop camera.

In that respect....if ML has actually managed to extract the full 14 stops of dynamic range from the 5D III and 7D (which, given that they are effectively doing two-frame HDR, I believe is highly likely...you have well more than 14 stops of original data to work with, and are only limited by the bit depth of the ADC), that means a 5D III with ML is actually capable of almost a stop more DR than the D800...at a lower native resolution.

 

[hambergler]

Dynamic range is Probably more Important to me than noise reduction or pixel count.
I think the way to do this is to split the sensor into highlights and shadows...
I rather have an 8mp 16-stop split sensor, than a 16mp one with less DR.
It can be done.
It is interesting that most HDR cameras take 3 exposures but nobody thought about changing the ISO Instead.
I believe in pushing things to the limit and opening up the possibilities of a device...
Like my jailbroken iPhone 5. Which is amazingly more useful and fun than the limited iOS from apple.


Kudos to magic lantern!
Keep up the great work!

why not 36Mp and the best DR ?

Not everyone can switch systems and if you do any significant video work whatsoever than 5DIII is several orders of magnitude better than the D800.

 

[Marsu42]

Fyi, don't know if this has been mentioned here: unlike other ml features this will *not* be ported to other cameras since only 5d3/7d have the required readout (for details see the pdf linked in the ml thread) :-\

 

[verysimplejason]

This is nothing real new and have been discussed at DPR years ago.

This document presents a sensor scanning trick that results in significant dynamic range im- provements, at the cost of vertical image resolution and aliasing, especially in highlights and shadows. The ISO (analog amplification) is alternated for every two scanlines, between two user-defined values (say ISO 100/1600), so half of the image is exposed for highlights and the other half is exposed for shadows.

Not sure what point you are trying to make. This feature hasn't been implemented yet. I would say that alone is new and newsworthy. A document released years ago doesn't do us photographers in the field any good...

+1. Theory is far from reality. Reality is the ML team is worthy the accolade their getting. I hope Canon takes a hint from this and apply it in a better way in the next generation of sensors.

 

[dtaylor]

why not 36Mp and the best DR ?

You seem to be lost. This is a CANON forum.

 

[dtaylor]

Here's hoping the ML team can get a final, stable release working on the 7D given the difficulties of dual processors.

I imagine both my 7D and my M will have ML on them sooner or later. DR gains on the 7D; focus peaking and intervalometer on the M. Canon should donate a lot of money to the ML team.

 

[Skulker]

This is nothing real new and have been discussed at DPR years ago.

This document presents a sensor scanning trick that results in significant dynamic range im- provements, at the cost of vertical image resolution and aliasing, especially in highlights and shadows. The ISO (analog amplification) is alternated for every two scanlines, between two user-defined values (say ISO 100/1600), so half of the image is exposed for highlights and the other half is exposed for shadows.

Not sure what point you are trying to make. This feature hasn't been implemented yet. I would say that alone is new and newsworthy. A document released years ago doesn't do us photographers in the field any good...

It's typical of him, just any chance to be negative and he's in there. He seems to think everyone should convert go a d800 and can't or won't see the advantages of anything with canon written on it.

 

[BozillaNZ]

It's good for video where output resolution will always be less than sensor's native resolution.

However for still photographs, you don't lose 2x the vertical resolution, you lose more. Since according to the technical document, it uses ISO 100 for sensor line 1,2, then ISO 1600 for line 3,4, and alternating like that.

It is how the hardware is wired up, two rows for each ADC/AMP. This has the effect of generating worse moire and alias on scenes with high frequency data since the spatial continuity is essential 1/4 that of original resolution.

 

[neuroanatomist]

why not, d800 has the same FWC as 1dx but 36 Mp
The best SLR sensor on the market today

I'm sorry, but that is a bold faced lie! The D800 has a full well capacity of 44k, while the 1D X has a full well capacity of 90k. That is a TWO FOLD difference between the two, bub!

Indeed. But let's not allow facts get in the way of the same restated DRivel from the DRoll DRones who DRool over DR. :

D800/E is already 36Mp with 14.5 stop DR... without the ISO tricks

You fell victim to one of the classic blunders - The most famous of which is "never get involved in a land war in Asia" - but only slightly less well-known is this: DxOMark's Scores are useless, biased Bovine Scat.

 

[neuroanatomist]

why not, d800 has the same FWC as 1dx but 36 Mp
The best SLR sensor on the market today

I'm sorry, but that is a bold faced lie! The D800 has a full well capacity of 44k, while the 1D X has a full well capacity of 90k. That is a TWO FOLD difference between the two, bub!

Indeed. But let's not allow facts get in the way of the same restated DRivel from the DRoll DRones who DRool over DR. :

D800/E is already 36Mp with 14.5 stop DR... without the ISO tricks

You fell victim to one of the classic blunders - The most famous of which is "never get involved in a land war in Asia" - but only slightly less well-known is this: DxOMark's Scores are useless, biased Bovine Scat.

from the whole sensor area you get exactly the same FWC from 1dx and d800

Full well capacity. You keep using that phrase. I do not think it means what you think it means.

From the whole sensor area? Tell me...what is a 'well' in FWC? How does the 'whole sensor area' affect the number of electons a well can hold?

You stated, "d800 has the same FWC as 1dx." Now you're stating, "from the whole sensor area you get exactly the same FWC from 1dx and d800." I was beginning to think you understood this stuff. I was wrong. See...unlike some, I can admit it. :

 

[jrista]

why not, d800 has the same FWC as 1dx but 36 Mp
The best SLR sensor on the market today

I'm sorry, but that is a bold faced lie! The D800 has a full well capacity of 44k, while the 1D X has a full well capacity of 90k. That is a TWO FOLD difference between the two, bub!

Indeed. But let's not allow facts get in the way of the same restated DRivel from the DRoll DRones who DRool over DR. :

D800/E is already 36Mp with 14.5 stop DR... without the ISO tricks

You fell victim to one of the classic blunders - The most famous of which is "never get involved in a land war in Asia" - but only slightly less well-known is this: DxOMark's Scores are useless, biased Bovine Scat.

from the whole sensor area you get exactly the same FWC from 1dx and d800

Full Well Capacity has to do with the physical pixel, nothing else. There is no "whole sensor area" when talking about FWC. Simply put, the maximum charge accumulation possible in the photodiode of each D800 pixel is 44972e-, while for each 1D X pixel it is 90367e-. There is no changing that, it is a fixed attribute of the hardware. If you downsample a digital image in post, you are normalizing noise across digital pixels, not physical pixels, and that is an entirely different process. Your numeric range is the same...either 8 bits or 16 bits, regardless of whether you have a D800 or 1D X, and for any levels above the noise threshold on the 1D X, the gains would be the same as for the D800. The primary gain with the D800 has to do with shadow levels, where you have more usable detail strait out of the camera than with the 1D X (however that still has nothing to do with FWC).

 

[neuroanatomist]

Whats matter is how many electrons can be hold before saturation/read out = and it is the the same from the two.

Mikael, does the D800 have the same FWC as the 1D X?

 

[neuroanatomist]

multiply d800 pixels with FWC and you get the same as 1dx

(a new question to discuss?)
then please keep your self to the subject, so it not will be a new head room or BSI/FSI discussion

Really? 1 pixel = 1 well. 1 well holds a given number of electrons. That number is different between the D800 and the 1D X. If you multiply the number of electrons that fill one well by the number of wells, you're not talking about FWC any more.

You want to discuss something else? You were the first to bring up FWC in this thread. Now that you've shown you don't understand the concept of FWC, by all means, let's move on. Even if you won't, I will. It's just not worth trying to explain to someone who refuses to admit when they're obviously wrong about a simple fact. :