Magic Lantern "Dual ISO" Function

Waiting on someone to incorporate the dual iso processing in the mac raw2dng, until then you can try using wine but I've never been able to get it to work (too stupid or too smart depending).

Nick Driftwoods 100/400 shot looks really good, I can crush the blacks without a big shadow detail penalty.

 
Tokpaler said:
I'm sorry if there's an obvious answer to this question, but why is the normal ISO 100 shot so noisy?
Click to expand...
From what I read it's because the actual RAW was very underexposed to protect the highlights and the shadows were brought back in post. He could have exposed to the right more to get less noisy darks, but then the highlights would be blown worse. I think it was more a test to show the noise difference in both when exposing for the highlights.
 
squig said:
Waiting on someone to incorporate the dual iso processing in the mac raw2dng, until then you can try using wine but I've never been able to get it to work (too stupid or too smart depending).

Nick Driftwoods 100/400 shot looks really good, I can crush the blacks without a big shadow detail penalty.

Click to expand...

Now this is a much better Dual ISO example. It shows a very noticeable improvement in both highlights and shadow detail.
 
What is the funky blue emerald artefact in the shot, you can see it clearly in between the leaves just above the building. It is visible in other places in the footage.
 
Yes, looking better. I think the emerald is related to the color on the dome but for some reason the dual ISO is going heavy on the saturation. Possible quirk.

One question, Do you think it is possible to lift the shadows in the ISO 100 sample to match the dual ISO shot? Seems like one could get that look without the dual ISO as these are RAW files.
 
Bassman2003 said:
Yes, looking better. I think the emerald is related to the color on the dome but for some reason the dual ISO is going heavy on the saturation. Possible quirk.

One question, Do you think it is possible to lift the shadows in the ISO 100 sample to match the dual ISO shot? Seems like one could get that look without the dual ISO as these are RAW files.
Click to expand...

It's not possible. That's what was done in the original sample and that's why the noise is so bad in the ISO 100 sample. The only other option to get the ISO shadows too look like the 1600 one is by using a higher ISO and killing your highlights. The point of the dual ISO is to get shadows like the ISO 1600 has and highlights like the ISO 100.
 
Comparing the two videos posted, it looks like the farther apart you ISO values are the more aliasing you get? Correct me if i'm wrong. But the details in the shadows on Nick Driftwood's video are pretty remarkable!
 
I've been pushing a1ex to go for a less aggressive setting from the beginning --> this 100-400 test looks a lot better! And with that the 5D3 still has 13.5 stops of usable DR: not bad at all :)

And yes, the bigger the difference between the ISOs, the more range that gets aliasing-moire: with 100-400 it's only the top two and the bottom two stops of light (2 aliased, 9.5 clean, 2 aliased), with 100-1600 it's the top four and the bottom four stops of light that get issues (4 aliased, 6+ clean, 4 aliased). Plus, when the ISO gets further apart, saturation and noise change too, so those central 6 stops of light get some minor artifacts too.

Early days, though: remember when the RAW thing came out and lots of people thought they'd never get the fps up to video standards? wait and be amazed
 
Last edited:
skital said:
It's not possible. That's what was done in the original sample and that's why the noise is so bad in the ISO 100 sample. The only other option to get the ISO shadows too look like the 1600 one is by using a higher ISO and killing your highlights. The point of the dual ISO is to get shadows like the ISO 1600 has and highlights like the ISO 100.
Click to expand...

I see your point but if that were a RAW photo I could go in Lightroom and lift the shadows without much of any penalty. Are the video RAWs that much less than the still RAWs for adjustment lattitude?
 
StMad said:
13.5 stops? I thought the base 5D had 10.7 stops...can someone explain the math to me?
Click to expand...

Read the title again. Also, look up HDR photography.

ML is using two exposures to make one higher dynamic range image.
 
I should have been more clear. If the base DR is 10.7 stops (right?), how does shooting 100/400 get us to 13.5 stops? Seems like a 2 stop gain.
 
Bassman2003 said:
I see your point but if that were a RAW photo I could go in Lightroom and lift the shadows without much of any penalty. Are the video RAWs that much less than the still RAWs for adjustment lattitude?
Click to expand...



The video RAWs are the same as stills. The 5D3 has a great ability to lift the shadows, but he said the ISO100 shot was very underexposed, which is why he got that much noise upon recovering the shadows. Remember that we don't know the exposure settings pre recovery. If you underexpose too much to protect the highlights in a high contrast scene not even the 5D3 can recover the shadows cleanly. The 5D3 does have a native dynamic range limit after all. The ISO 100 was intentionally underexposed to protect highlights so that the ISO 1600 could give the clean shadows while retaining the highlights. Your option would have had more highlight clipping to retain the clean shadows of this particular scene.
 
Last edited:
StMad said:
13.5 stops? I thought the base 5D had 10.7 stops...can someone explain the math to me?
Click to expand...

Combining the 2 ISOs increases the dynamic range. If you you want a better explanation read a1ex's pdf or google HDR photography as James already suggested.
 
Nah, that's cool...I simply got the base iso wrong. It'd be interesting to see a side by side with the BMC. Anyway, the mk3 (credit to ML) seems to be the camera that keeps giving.
 
You must log in or register to reply here.
Back
Top