Why I think Knee and Slope settings are the Devil

Michael Solomon said:
Nirv, which profile is that?
Click to expand...

My own.

I'm not entirely convinced that the profile is to blame nor the camera.

I have a suspicion that some NLEs are not handling the avchd files correctly. The reason why I say that is because I've tried all sorts of loony things to recreated some of the halos and banding and can't.

It could also be that some people have been playing with the colour matrix to such a degree that when the luminosity approaches over exposure that unwanted posterisation effects occur.

Is there any way I can find out what version of firmware I have? Any ninja button combinations to bring up that info?
 
Last edited:
Nirv said:
My own.

I'm not entirely convinced that the profile is to blame nor the camera.

I have a suspicion that some NLEs are not handling the avchd files correctly. The reason why I say that is because I've tried all sorts of loony things to recreated some of the halos and banding and can't.

It could also be that some people have been playing with the colour matrix to such a degree that when the luminosity approaches over exposure that unwanted posterisation effects occur.

Is there any way I can find out what version of firmware I have? Any ninja button combinations to bring up that info?
Click to expand...

I am seeing similar results inside and outside of multiple NLE and viewers. I use VLC for just looking at MTS clips.

Here is another example for you. Same shot, only difference is I cranked up the gain on the FS100. The one on the right, which is blown out, looks great in terms of highlight handling and similar to your results. The one on the left, you can see the banding. Now, obviously this is NOT a good shot, but just one that illustrates that the banding is not consistent. During exposed shots that normally look good on other camera, this is the sort of thing that pops up. Like I said, look at the sky on that FS700 video from the "4 days" shoot, that yellow is horrid to me.

Screen shot 2012-04-18 at 2.48.26 PM.jpg

There are MUCH more examples of this online you can find. One of the perfect examples is getting yellow shift in highlights on hair where the highlight is allowed to blow out. And, with waves at the beach, getting cyan in the foam.

If you've found a way to eliminate this issue, please, share your settings! :)
 
Hmmm, you've got me wondering something. Is this an issue with material over IRE100 and how it's maybe being handled? What does the software do when the range exceeds a certain range. Maybe these color shifts are being caused in how the top end by default is handled and because the curve extends up into > 100, that it's getting a bad side effect from it, if that makes sense.
 
I am uploading the original file off the camera to Vimeo.

The file is undoctored and as such the Vimeo streaming version may clip the highlights so please download the original.

It might be worth people testing it in their NLE of choice so we can rule out the NLE as a source of problems.

Uploading is taking a while and I'll edit this post and add a URL when done.
 
Rick Burnett said:
Hmmm, you've got me wondering something. Is this an issue with material over IRE100 and how it's maybe being handled? What does the software do when the range exceeds a certain range. Maybe these color shifts are being caused in how the top end by default is handled and because the curve extends up into > 100, that it's getting a bad side effect from it, if that makes sense.
Click to expand...

Exactly...

FWIW, I am using the much maligned FCPX which handles everything I have ever thrown at it in an exemplary way.
 
seven.b said:
Are these highlights considered 'bad' to you guys? This was shot on the FS100, but I'm just trying to wrap my head around this highlight drama.

https://vimeo.com/35609128
Click to expand...

I think that is the jist of it......everyone has there own thoughts on what's an acceptable anomoly for each cam.
personally, I liked that vid. then again, I like an occasional lens flare and I like the way bokeh gets stretched vertically with certain lens' but that's just me....
 
Here are some shots from the "4 days" video with the FS700. Note, early on, with the dog running around, the blow-out of that sun looks good. However, these are what I have issue with. The thing is, some of the shots look nice, even from these same segments. Just a certain intensity above some point and it starts doing things I am not a fan of (hue shifting).

Screen shot 2012-04-18 at 4.26.54 PM.jpgScreen shot 2012-04-18 at 4.22.56 PM.jpg

And this is something strange, but I think he sped up the footage and the NLE did something to the footage as I've not seen this in anything else.

Screen shot 2012-04-18 at 4.19.58 PM.jpg

Don't get me wrong, almost all of this video is AWESOME. Andrew did some great shots. It's the ones above I just do not like personally.
 

Attachments

  • Screen shot 2012-04-18 at 4.21.32 PM.jpg
    Screen shot 2012-04-18 at 4.21.32 PM.jpg
    40.4 KB · Views: 0
Rick Burnett said:
And this is something strange, but I think he sped up the footage and the NLE did something to the footage as I've not seen this in anything else.

View attachment 52158

Don't get me wrong, almost all of this video is AWESOME. Andrew did some great shots. It's the ones above I just do not like personally.
Click to expand...

Frame blending, the only one that looks bad to me is the marshmallow
 
Having just watched the Apprentice and seen similar halos around bright lights. The pragmatist in me would say put this to bed. The Apprentice is not shot on FS100s so it's obviously a Sony setup.

If you are still bothered by this look in previous threads where I discuss the 709 curve and colour mode setting.

Case closed.
 
Sorry, it's not case closed. Maybe for you, but for me I still deal with it and until I can find easier ways to work around it, I am still going to explore what to do with the camera to get around it, or, hopefully get some better correlation to what I want with the FS700 as Adam Wilt is seeing.

If you read everything I wrote, you'd see that I stated other cameras get a color shift as well, in particular I've looked closely at the Alexa. The difference is, I NEVER see a hard switch like what you see on the blue LED I posted. THAT is what I want to eliminate. With the FS700 footage, so far, those hard clips seem to be a bit more minimized, however, the color shifts can still be pretty bad, as I've shown from the examples. I can live with a small colorshift area that eases in and out with a nice gradient. That doesn't always happen with my FS100. Sorry.
 
Postmaster said:
I posted this in a separate tread, just saw your guys are already over it:



Knee setting: 105% and slope +5 in combination with Cinegama 2 does a nice job here.

You pay for it with lowlight sensitivity and some noise.
But you don't have to shy away from highlights anymore and when you expose correctly, it´s fine.

Try this out if you want:

Black Level (variable - start at 0 and adjust to your scene)
Gamma: Cinetone2
Black Gamma High +7
Knee: 105% and slope +5
Colormode: Cinematone2 +8
Color Level -2
Colorphase 0
Depth: R+4, G+7, B+7, C+7, M+4, Y+7
WB-Shift R-B, R-7, B0
Detail: Level 0, Manual ON, BWType3, Limit4, Chrispening5,


Frank
Click to expand...

These setting definitely reduce the banding on the LED adapter, and I think they look really good. The red is still reacting non-linearly I feel in the shots I tested, and I think these setting reduce the saturation in the color shift which pulls down the effect of the hue shift. I tested shooting a bunch of different colors, much better.

My metabones adapter came in today and I have to say, I'm a little annoyed that they couldn't just make the hole circular in front of the sensor. With a F1.4 lens, it's blocking the light coming in depending on the angle and causing the normally circular bokah to have the tops and bottoms clipped off a bit.

As soon as I get the new firmware installed, I am going to do some more tests. With the expanded focus, I can see the banding now, with the 4x and 8x I can see it better. (This also tells me the problem is not the NLE because, as said, I see it in the onboard monitor). I'll post a picture of what I get in the end with Frank's settings to compare with what I posted earlier.

*THIS* is what I love about the FS100 forum, a lot of people working together to solve issues. Even if we don't always agree! :P
 
Rick Burnett said:
*THIS* is what I love about the FS100 forum, a lot of people working together to solve issues. Even if we don't always agree! :P
Click to expand...

Agreed. I have been following this thread VERY closely and I want to thank all of ya'll for your posts. I have learned so much about the FS100 from you great guys. Thanks.
 
So.... I know there's been a raging throbbing debate the past week regarding a 105% knee.

Well, I decided to go that route for a narrative short that I shot this past weekend. I used the ABNorm profile except changed the slope to +1 and the knee to 105%

All in all, I'm happy with how everything came out. It required grading (I find abnorm sort of "neutral"). But the final images were pleasing and the client is happy.

They asked me to do a quick speed grade of a couple clips so they could get an idea of how the final product may look before they took it to their go-to colorist.

You can see it here: https://vimeo.com/40627723

Much happier with how blown highlights as well as how 80-100 IRE was handled with the higher knee.
 
Michael Solomon said:
So.... I know there's been a raging throbbing debate the past week regarding a 105% knee.

Well, I decided to go that route for a narrative short that I shot this past weekend. I used the ABNorm profile except changed the slope to +1 and the knee to 105%

All in all, I'm happy with how everything came out. It required grading (I find abnorm sort of "neutral"). But the final images were pleasing and the client is happy.

They asked me to do a quick speed grade of a couple clips so they could get an idea of how the final product may look before they took it to their go-to colorist.

You can see it here: https://vimeo.com/40627723

Much happier with how blown highlights as well as how 80-100 IRE was handled with the higher knee.
Click to expand...

Nothing popped up HL wise that I could see, I even watched with sound off so I wouldn't be tempted to pay attention to it. Looked great. Was that red pillow really that red? Just have to ask. :)

If I were to make any comment, it might be on the dynamic range of the image with her singing and playing guitar. The detail in her hair is missing. Of course, a light could fix that, just something that popped out to me. What I really liked is the guitar strings didn't alias over the sound hole, which is something that happens VERY easily.

In any case, thanks for sharing. I think I am a believer of higher knee myself at this point. So far with my testing, still much happier. Still pushing it though.
 
I was just reading through this thread and realized something. About 6 or 7 months ago, I slightly modified one of Frank's
picture profiles and have been using it quite a bit. I just looked at it and it has the knee point set at 102.5 and
the slope at +1. So maybe I was onto something way back then. Of course, i can't really take much credit for it,
it was mostly Frank's profile and I tweaked a couple things and just liked the way it looked. So not at 105 and +5,
but closer to that direction than some of the 'aggresive' knee point and slopes that are sometimes used.
 
I just updated the firmware, and call me crazy, but for some reason I feel like the cinegammas are giving off better hilight control. So much so that I'm prefering the look of cinegamma 1 now. From what I remember the FS100 was originally rated at a base ISO of 500. Now you can go down to ISO 400 with cinegamma 2, but it's restricted when cinegamma 1 is enabled. I wonder why?
 
You must log in or register to reply here.
Back
Top