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 :)

ISO... today

Status
Not open for further replies.
Jarred Land said:
I am actually in Chicago right now on Wanted.. but ask that question again in a day or two and it's likely to change.
Click to expand...

Say hi to Angelina for me. If she's interested you can pass along my phone number or some modeling pics that I shot in front of my bathroom mirror. :weight_lift:

...jus' sayin.
 
Has anyone heard anything about analogue gain adjustment?
 
As in adding electrical charge to the sensor? I think the point of the approach Red has taken to Gain is to avoid such permanent actions. I'm guesing the difference in quality is neglidgable or they would have implemented it.

BTW this is always what I thought would happen, ISO 500 just made sense if you're protecting highlights. Way to go Red team!
 
An effective "Negative analogue gain" would be possible only if the normal or default condition is using some positive analog gain prior to A/D conversion, and that would also mean that some potential dynamic range is being left on the table. As someone else mentioned, if your image is giving you clipping in one or more channels, no subsequent electronics can bring that back, no matter what gain setting is dialed in.
 
So it might be a good idea to slightly underexpose the RAW all the time?
 
I think the Red team has looked at this... you can see the results in the various ISO threads. They shot at various levels of (analog) gain. You can compare between using analog gain and electronic/digital gain. To do the latter, throw the image into photoshop and go to work with the levels adjustment layer.

2-
As someone else mentioned, if your image is giving you clipping in one or more channels, no subsequent electronics can bring that back, no matter what gain setting is dialed in.
Click to expand...
You can sort of do this. If the white balance is off, then not all the channels clip at the same time.

You can do some highlight recovery techniques to guess the missing channel information... it's the same idea as highlight recovery in the RAW tool in Photoshop.
 
500 asa is great news. That means 1000 asa is only one pushed stop away.
NDs are WAY cheaper then generators and large HMIs
Nice to know you could get away with lighting a night exterior with a 4K instead of an 18K if you had to.

Crystal synch put-puts anyone?
 
Did I mention this is my double-ton post?

Sorry lame...:tongue:
 
GlennChan said:
You can do some highlight recovery techniques to guess the missing channel information... it's the same idea as highlight recovery in the RAW tool in Photoshop.
Click to expand...
Lots of well-informed folks on this board :) You can extrapolate some missing detail from the other channels, that is true; I've done some of that in Photoshop also. Still I would consider that more a "disaster recovery" technique rather than something you'd actually plan your exposure around.
 
Gibby said:
Yes, there is a Gain adjustment measured in 3dB steps available. That affects the RGB recordings, video outputs and EVF feed. When recording RAW the value is stored as metadata, so you can use this value for gain in post, or ignore it - your choice.

Stuart English
7/1/07
RED User
http://www.reduser.net/forum/showthread.php?t=3039&page=2

...

There's two things - analogue sensor gain ahead of the A-to-D, and digital gain, which occurs post digitisation. Both are useful.

What we're doing in these tests is to show that if you under expose, and gain up in post (ie digital gain) what results you get.

Graeme Nattress
7/17/07
RED User
http://www.reduser.net/forum/showthread.php?t=3235&page=11
Click to expand...


Wow, Gibby, thanks for that in-depth reply. So, if I understand correctly, when recording REDCODE Raw, no matter what you set the gain at, the sensor is going to capture at a sensitivity of 500ASA. If you under expose, you adjust later in post. Any gain setting is just metadata that can be applied later, right?

This would make sense since the CMOS based Phantom acts the same way. In fact, white balance is also stored as metadata too (on the Phantom)

So, by adjusting gain, you can't force the camera to be less light sensitive and substitute the adjustment for ND filters --which would be handy--.

But, I am having a harder time understanding the differences when recording RGB. Maybe it's because it's late and my mind is fried.

Thanks again,
Phil

www.artbeats.com
 
Gibby said:
Yes, there is a Gain adjustment measured in 3dB steps available. That affects the RGB recordings, video outputs and EVF feed. When recording RAW the value is stored as metadata, so you can use this value for gain in post, or ignore it - your choice.
Click to expand...

Ok so let me just clarify something. What I am understanding from these posts (thanks Gibby) is that no matter where you set the gain, the data recorded in a RAW file will always be exactly the same. All that is changing is the metadata. When you get to REDCine or FCP the metadata will adjust the apparent exposure that you see, but again RAW data is never transformed. Am I right so far?

Does the EVF and LCD reflect the gain adjustment while shooting?
Does the HD-SDI?

It seems at first like there is an advantage to using analog gain adjustments, despite the increase in noise. But I guess this all comes from a misunderstanding of the Mysterium sensor. Can you explain the difference between a RAW image where the sensor gain is adjusted in post, and one where that gain is adjusted analog?

IBloom
 
Thanks Steve,
and also thanks Graeme and Stuart.

500 is a great place to start in my experience...

Just a quick question: I understand that the digital gain doesn't affect the Raw image, but only adjust visualization in the VF or Video Out, but what happen if I need to see an Histogram (just to check my curve, floor and ceiling), I have any opportunities?

A sort of curve-visualization in the VF (maybe with/without gain applied)?

Just a curiosity, thanks...
 
ISO 500 sounds wonderful.

To which extent the wavelet compression is in our favour in low light shooting? I mean, noise is typically about high frequency components although easily observed by eye. As the compression employed is lossy, thus some data is discarded, and one could easily come to the conclusion that what is thrown away is especially about noise.

Does this partly explain the wonderful preliminary RED test shots up to ISO 4000? Furthermore, does the signal processing on the sensor level raw data yield more "space" compared to filtering out the noise in post from a RGB-file or from a film?
 
Technical question, if 500ASA is the native ISO of the sensor and we cannot go lower than this, how is thie achieved in a DSLR for example?

I can choose anywhere from 100 to 1600 on my DSLR, the result being at 100ASA I can shoot at lower shutter speeds outdoors than at 400 or 500ASA . It seems, though it could be a trick, that at 100 you need more light coming into the camera. Are they not dialing back their sensor here too?

Or is it that my DSLR is rated at 100ASA natively and we're increasing gain from there up? Red on the other hand is rated at 500ASA so it will only start at 500 and up?
 
jlaurili said:
So it might be a good idea to slightly underexpose the RAW all the time?
Click to expand...

When shooting RAW with a DSLR, I believe it is more usual to actually do the opposite and "expose to the right" (of the histogram). So long as the highlights are not blown, it's arguably better to darken the image in post than lighten it and increase noise.
 
Status
Not open for further replies.
Back
Top