RAW S-Log Clip to Download

[Doug Jensen]

Originally Posted by Bruce Schultz

Roger all of that Mr, Jensen, but I was directing that suggestion to Mr. Villalba who started this thread - you both have similar first names.

Sorry, my mistake.
Ever since I created the 126 test clips I referred to in post #18, I have been deluged by requests from people to do further testing as well. I mistakenly made the assumption you were suggesting the same thing. Sorry.

[Douglas Villalba]

Originally Posted by Alister Chapman

That's not correct this is log and 10-87% equates to 8 to 9 stops.

The S-Log curve is as follows:

0-10% 3 stops (If trying to use 8 bit this means each stop represented by only 8 grey shades, 10 bit is 30 grey shades per stop)
0-20% 5.5 stops
0-30% 7 stops
Middle Grey 38%
0-40% 7.5 stops
0-50% 8.5 stops
0-60% 9.5 stops
0-70% 10 stops
0-80% 11 stops
0-90% 12 stops
0-100% 12.6 stops
0-104% 13 stops

Arguably the first stop above black is going to be quite noisy so is unlikely to be of any significant use in a real world image, so measured dynamic range is 13 stops but latitude 12 stops.

That is why I did a test that is useful to me.

My brain can't compute that for example from 40% up every 10% increase is 1 full stop then from 60-70% it is only 1/2 stop and then from 70-80% it goes back to 1 full stop all the way to 90%. My brain can compute a Log slope, but a sharp drop in the middle middle, I can't.

Can't get any easier to see what an f-stop difference does than opening and closing the iris one stop at a time.
You can stop trying to convince me to trust your theory. It will never happen.

[Alister Chapman]

The anomaly around 60-70% is due to rounding of the exposure to the nearest 0.5 stop. I suggest you take a look at any of the published curves or S-Log curve plots on the web and make your own assessment of the exact numbers.

S-log curve image by hingsberg on Photobucket

It's not theory it's fact. Fact that is borne out by every other correctly done S-Log measurement that's been done. I suggest you read the Sony S-log white paper or watch the Abel Cine tests etc.
If you don't like the facts that's your choice but don't keep telling me I am wrong and only your test is right. I don't agree with your test results of a 9 stop range or that 10-87% is only 4-5 stops because that's not the case and that isn't opinion or speculation, it's stone cold fact.

Your test obviously has value to you, it may have value to others as well, your exposure method is obviously one you like, but statements such as 10-87% only being 4-5 stops show that you do not fully understand the curve and it's exposure range, either that or you are talking a different language to everyone else. In that case don't expect us to understand you.

[Douglas Villalba]

Originally Posted by Alister Chapman

It's not theory it's fact. Fact that is borne out by every other correctly done S-Log measurement that's been done. I suggest you read the Sony S-log white paper or watch the Abel Cine tests etc.

I understand Theory like this.
Wilkipidia: "Theories are analytical tools for understanding, explaining, and making predictions about a given subject matter."
In order for ME to be fact, you have to show me an actual situation (Zacuto shootout scene video) with a range over 7 stops. Less than that I find S-Log totally useless and just an extra step that 709 can handle even at 8 bits 4:2:0 straight out of the camera.

If you don't like the facts that's your choice but don't keep telling me I am wrong and only your test is right.

I don't remember saying that you are wrong or that my test is the only test that is right. I said and I repeat, My test is the only one I have seen that shows ME the facts that I wanted to see.
I have only use your name to praise the work that you have done.

I don't agree with your test results of a 9 stop range or that 10-87% is only 4-5 stops because that's not the case and that isn't opinion or speculation, it's stone cold fact.

I shouldn't even call what I did a test. All I did was to shoot a video that SHOWS ME what happens to the under or under expose areas in 1 F-STOP increment.

statements such as 10-87% only being 4-5 stops show that you do not fully understand the curve and it's exposure range, either that or you are talking a different language to everyone else. In that case don't expect us to understand you.

The 10-87% came from just dropping your clip on the Apple Color Timeline before any grading. I am sure if you do the same you should get the same result unless different waveforms read differently. I only use the waveform in post to make sure that my whites don't go over 100% and my blacks don't go bellow 0%.
When I said that the clip was in the range 4-5 f-stops, I guessed. I don't know how anyone could use a waveform to determine accurately f-stops. I used experience on similar lighting situation and the Sunny 16 Rule. Again I was just asking you if you had a one frame clip of a 13+ f-stops shot.

I thought of doing one with a single light until I did the math.

I order to do it I had to place the 1st object 3' from the light...
3' = 0 f-stop under
6' = 1 f-stop under
12' = 2 f-stop under
24' = 3 f-stop under
48' = 4 f-stop under
96' = 5 f-stop under
192' = 6 f-stop under
I can only do 5 f-stops in my living room ;-)

[Alister Chapman]

If your guessing about exposure ranges then don't make statements:

Originally Posted by Douglas Villalba

The clip for download is in the 4-5 f-stops (10%-87% waveform). Do you have any single frame DPX frame with a range over 0-100%?

It's fine to guess or estimate, I haven't got a problem with that, I do it. But you stated the range was 4-5 stops and it felt like a criticism that somehow I was trying to short change people with a restricted dynamic range clip. I suspect your waveform readings are correct, and that makes it 8 to 9 stops.

If you know the gamma you are using then of course you can use a waveform monitor to measure latitude or F-stops. As I tried to explain earlier a video camera is no different to a light meter. You have a sensor and a voltage output. By measuring the output you can figure out the amount of light falling on the sensor or as it has a lens different parts of the sensor. That's all a light meter does. It has a sensor and you measure the voltage output of the sensor and from that calculate the the light levels. So if you know the characteristics of the gamma curve you can very accurately measure the exposure in f-stops. With a waveform monitor not only can you measure the average levels but the minimum and peak levels, all at the same time at any point in the image, something simply not possible with a light meter.

Frankly that's something I rarely do other than when testing a camera. Once I know the characteristics of the curve and how best to use it all I need to do is ensure my exposure range falls in the correct voltage range using either the spot meter built into the F3 or a waveform monitor. Most half decent monitors have a built in waveform monitor these days. Certainly my TV Logic and Transvideo's do. The waveform monitor is the best tool for setting accurate exposure with a video camera. Not always practical, especially for run and gun, but a waveform monitor is much more accurate than a light meter for measuring actual exposure.



The scene in the new Zacuto shootout has a 12 stop range.

[Douglas Villalba]

Originally Posted by Alister Chapman

If your guessing about exposure ranges then don't make statements:


It's fine to guess or estimate, I haven't got a problem with that, I do it. But you stated the range was 4-5 stops and it felt like a criticism that somehow I was trying to short change people with a restricted dynamic range clip. I suspect your waveform readings are correct, and that makes it 8 to 9 stops.

If you know the gamma you are using then of course you can use a waveform monitor to measure latitude or F-stops. As I tried to explain earlier a video camera is no different to a light meter. You have a sensor and a voltage output. By measuring the output you can figure out the amount of light falling on the sensor or as it has a lens different parts of the sensor. That's all a light meter does. It has a sensor and you measure the voltage output of the sensor and from that calculate the the light levels. So if you know the characteristics of the gamma curve you can very accurately measure the exposure in f-stops. With a waveform monitor not only can you measure the average levels but the minimum and peak levels, all at the same time at any point in the image, something simply not possible with a light meter.

Frankly that's something I rarely do other than when testing a camera. Once I know the characteristics of the curve and how best to use it all I need to do is ensure my exposure range falls in the correct voltage range using either the spot meter built into the F3 or a waveform monitor. Most half decent monitors have a built in waveform monitor these days. Certainly my TV Logic and Transvideo's do. The waveform monitor is the best tool for setting accurate exposure with a video camera. Not always practical, especially for run and gun, but a waveform monitor is much more accurate than a light meter for measuring actual exposure.



The scene in the new Zacuto shootout has a 12 stop range.

I hope that you understand that my "test" is not about how to measure light or reflection. Your way or mine. It just shows what the camera can register when light hits the censor at different intensity in a 1 f-stop variations. I even used a still lens to get the full click in between f-stops.

The discussion we have about light meters and waveform has nothing to do with my test.
We will never agree on this because I am interested on light landing on the subject regardless of its reflectance value and you are interested on the light reflecting off of the subject. They are both useful, just different and we each feel more comfortable with our different tools.

In Andy's test the light landing on the chart is even. Therefor there is no difference in exposure on the chart. The same amount of light is landing on that black as it is landing on the whites and all the way through.
Now the reflectance of chart varies in the shades from black to white and that is what the waveform is seeing.
Are you still with me on this?

[Alister Chapman]

Do you really think I don't understand the difference between incident and reflected light or how a grey scale chart works. Don't insult me. Andy's tests were done with a transmissive chart by the way.

If you think that waveforms are irrelevant when measuring exposure or assessing latitude then I'm leaving this now pointless discussion.

[Douglas Villalba]

Originally Posted by Alister Chapman

I'm leaving this now pointless discussion.

Good idea.