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 :)

Playing with the Forged Carbon Fiber Weapon and it's AWESOME!

Just seems that High Shadow Noise and High DR are mutually exclusive. To me, the stops of DR don't start getting counted until there is usable image value in the shadows and stop getting counted when highlight clipping first occurs. Which is about 9-10 stops in the test video. Not sure how we can get to 14-16 stops from there (which is mathematically way different) on any popular camera currently out there other than it's just marketing hyperbole.
 
Zeb B said:
To me, the stops of DR don't start getting counted until there is usable image value in the shadows and stop getting counted when highlight clipping first occurs.
Click to expand...

On the top end, it is easy to see when it clips. The problem on the bottom end is that people don't agree on what is usable. Some like the texture of the noise and consider it okay while others do not. There is no published standard on what is acceptable. There is no published standard on how to measure a cameras dynamic range just as there is no standard on how to measure the contrast ratio of a display.

Two examples are the Xyla-21 backlit chart, which is series of gray patches where each one is twice as bright as the one before it, going from left to right. Then there are real world tests where you +/- n stops from what is considered proper exposure, which is what Geoff uses in the CML camera tests. One issue with the backlit chart, besides being synthetic, is that it is at one color temp. I believe it is around 4200 kelvins. David told me once, but I don't recall the number off hand.

You need to test cameras at different color temps and I would go so far as to say different light sources. I think for LED, you need to agree on one particular full spectrum LED with a known quality. There are too many poor quality LED lights. But then I guess you could also pick out a poor quality LED and throw it in the mix as well. It's a rather large test matrix. :)
 
I do agree that Dragon, then Weapon, is one of the best (if not the best) in the biz for DR. Just seems strange there isn't a more straight-up way to measure true usable DR stops, like a standardized S/N base floor in the shadows, to which all cameras start their DR stop count from.
 
Zeb B said:
Just seems that High Shadow Noise and High DR are mutually exclusive. To me, the stops of DR don't start getting counted until there is usable image value in the shadows and stop getting counted when highlight clipping first occurs. Which is about 9-10 stops in the test video. Not sure how we can get to 14-16 stops from there (which is mathematically way different) on any popular camera currently out there other than it's just marketing hyperbole.
Click to expand...

Correct me if Im wrong, but shouldnt we also be factoring the existing DR in the ORIGINAL frame? Surely there should be quite a bit there? Otherwise by this logic the F3 has 8 stops DR and Alexa 11?
 
When measuring the total captured dynamic range of any camera system it's best to measure and read individual stops. Even the Xyla can produce "contamination", though it's the best one banger chart out there. The question for all is how much of that can be considered "usable stops". As you dig closer to the noise floor is where you'll see more of the image noise/texture/grain.
 
Nick Morrison said:
Early tests suggest it could be more..3200 ISO looked CLEAN....on Weapon.
Click to expand...

Those tests mean nothing.

There's no point in assuming anything until we can do a proper test ourselves. Half a stop sounds reasonable. It's the same sensor with better A/D conversion.
 
Phil Holland said:
When measuring the total captured dynamic range of any camera system it's best to measure and read individual stops. Even the Xyla can produce "contamination", though it's the best one banger chart out there. The question for all is how much of that can be considered "usable stops". As you dig closer to the noise floor is where you'll see more of the image noise/texture/grain.
Click to expand...

That's why it's strange there isn't an industry standard on where the shadow floor is (using S/N) to start the count. That and what is considered clipping; 1 pixel or 10%+ of the total frame? Without standards the whole DR claim topic seems an exercise in marketing masturbation. From the above Vimeo test it seems that we got about 9 stops, with 5 under and 4 over the "Key Side Metered" area before loss of color fidelity and clip.

To educate my simple mind - what's wrong with interchanging steps of EV with stops of DR in this example?
 
Zeb B said:
That's why it's strange there isn't an industry standard on where the shadow floor is (using S/N) to start the count. That and what is considered clipping; 1 pixel or 10%+ of the total frame? Without standards the whole DR claim topic seems an exercise in marketing masturbation. From the above Vimeo test it seems that we got about 9 stops, with 5 under and 4 over the "Key Side Metered" area before loss of color fidelity and clip.

To educate my simple mind - what's wrong with interchanging steps of EV with stops of DR in this example?
Click to expand...

I measure with light, but since S/N ratio can vary on a per sensor basis I don't know if that's what's most important for our needs as shooters. I only care about what the camera can "see", if it's accurate, and where the limits are. For me as long as you can see a separation between values and they appear as a full stop (or a measurable 1/3rd or whatever at the limit) that's what I'm interested in. As long as you are measuring each stop and "know" it's a full stop you can use EV I imagine. Which Vimeo video are you referencing?
 
Zeb B said:
This: http://www.reduser.net/forum/showth...t-s-AWESOME!&p=1535009&viewfull=1#post1535009
Click to expand...

-5 stops corrected still shows details so it's not a DR test. When your neutral gray chart (once corrected) don't show any difference with the surrounding darker area (when under exposing) and once your neutral gray chart don't show any difference with the brighter part of the image (when over exposing) it's where your DR stops.

It's not 9 stops which is close to a 5d DR in video mode!
 
Patrick Tresch said:
-5 stops corrected still shows details so it's not a DR test. When your neutral gray chart (once corrected) don't show any difference with the surrounding darker area (when under exposing) and once your neutral gray chart don't show any difference with the brighter part of the image (when over exposing) it's where your DR stops.

It's not 9 stops which is close to a 5d DR in video mode!
Click to expand...

I get it's not a DR test. Point is: a camera can test well in the lab with DR charts but suck ass in real world Useable Stops since there is whacked color fidelity and distracting noise at the edges of performance. The shape of that bell curve is not measured with a DR chart test. Can there not be a better, standardized measure of DR? Like one using S/N on the low end and % of clip limit and color fidelity on the high end?

Saying that the shadow chroma noise is "filmic and organic" is wankery. When film is underexposed it doesn't puke RGB garbage - it's just black ;-)
 
Phil Holland said:
When measuring the total captured dynamic range of any camera system it's best to measure and read individual stops.
Click to expand...

You can actually measure one stop at a time with the Xyla. It has two inserts that slide into the sides to block out the steps you do not want to see or measure.
 
Zeb B said:
Saying that the shadow chroma noise is "filmic and organic" is wankery. When film is underexposed it doesn't puke RGB garbage - it's just black ;-)
Click to expand...



Screen_Shot_2015_08_17_at_6_40_51_PM.png
 
You must log in or register to reply here.
Back
Top