More Neon Overexposure Issues

[Dermot Shane]

"in ACES wiht the ODT set to bypassed/none, this shot shows the red chanel being clipped" - I don't think that's a valid way to determine clipped data. Looking at the raw data or my software version of the in-camera raw exposure check shows the red channel is not clipped. Looking at the brightest section I see raw 16bit code values of RAW 51749,14596,1236 (16) which is indeed very hot, but not actually clipped.
Graeme

interesting, i used the DCTL that Scott Dyer created for ArriRAW, applied in the media pool so it's AP0, usualy that done a pretty good job of killing off neon light damage in any of your camera's as well, but i didn't go any further, and i did see red channel as clipped after winding down exposure a country mile or so... will try again with the highlights de-sated to near 0...

And i didn't try the old skool matrix that i've used to good effect up to the point that i started using Scott's DCTL, the DCTL is certanly easier to apply

thanks for the reason to take a second look,

 

[rand thompson]

Hrvoje,

I really like the way that looks!

I tried again using REC709-BT1886

[/url]Capture by rand thompson, on Flickr[/IMG]

1) As per Aaron and Graeme, I desaturated the Reds to 37-Sat, but worked mostly with the Yellow/Orange discoloration again

[/url]Capture 2 by rand thompson, on Flickr[/IMG]

[/url]REC709-BT1886 by rand thompson, on Flickr[/IMG]


2) Used a "3d Qualifier" isolated The Yellow/Orange problem area while masking the Red color

[/url]Capture 3 by rand thompson, on Flickr[/IMG]

Capture 5 by rand thompson, on Flickr


3) And the final image.

[/url]Capture 6 by rand thompson, on Flickr[/IMG]


Final image stayed Legal in the vectorscope for the Red channel.

[/url]REC709-BT1886 by rand thompson, on Flickr[/IMG]

 

[Hrvoje Simic]

Thanks Rand.

interesting, i used the DCTL that Scott Dyer created for ArriRAW, applied in the media pool so it's AP0, usualy that done a pretty good job of killing off neon light damage in any of your camera's as well, but i didn't go any further, and i did see red channel as clipped after winding down exposure a country mile or so... will try again with the highlights de-sated to near 0...

And i didn't try the old skool matrix that i've used to good effect up to the point that i started using Scott's DCTL, the DCTL is certanly easier to apply

thanks for the reason to take a second look,

Did you check the image posted before ?

 

[Dermot Shane]

Did you check the image posted before ?

Yup, many images, mostly using some sort of quaifiier, keyer, curves etc... but those tools are not a first choice for me in this case

Your looks best by far, and far more likely to work with shots that have skintones lit by neon, something that the keyer soultion often fall on it's face with... but this shot is ony a neon sign, no skintones, so yea, not a hard test, and dead easy to sort with keyers.

Check out post #4, it shows what i was seeing (although with a 709 ODT, pretty much the same, but more spread out without a ODT)...

I'll give oit a look with a "neon light damage" powergrade i have hanging around, it may be better at this than the Acadmy DCTL..

So far the DCTL has been working really well with any camera, and simpler than messing with the 3x3

And one size does not fit all, so for me the takeaway is to look further if the DCTL does not kill off neonlightcrap...

 

[Hrvoje Simic]

yup,many images, mostly using some sort of quaifiier, keyer, curves etc...

check out post #4, it shows what i was seeing (although with a 709 ODT, pretty much the same, but more spread out without a ODT)...

i'll give a shot with a "neon light damage" powergrade i have hanging around, it may be better at this than the Acadmy DCTL..

so far the DCTL has been working really well with any camera, and simpler than messing with the 3x3 & highlight sat, but it may have it's limits and one size does not fit all, so for me the takeaway is to look further if the DCTL does not kill off neonlightcrap...

Did you check the image in post #19 ?

No qualifiers, keyers or anything similar dismantling the image.
Graeme's Log3G10 RWG + custom tool on top of that.

One transform cannot fix many cameras, for optimal performance it has to be designed per each sensor's unique response.

 

[Dermot Shane]

Did you check the image in post #19 ?

No qualifiers, keyers or anything similar dismantling the image.
Graeme's Log3G10 RWG + custom tool on top of that.

One transform cannot fix many cameras, for optimal performance it has to be designed per each sensor's unique response.

yea i did, as mentioned, your resualts look great to me

no keyers is a good thing once skintones lit by LED's get into the mix

i can get to the same place and keep working in ACES, i do this a fair bit.. police drama / led lights on cop cars / actor's faces lit by the same...

using the DCTL has sorted this with Epic, f55, Dragon and Alexa with a one button push, and that helps alot in getting through a show...

i do have tools to sort this that i made myself, and they will work up to the point that the camera clips - and keeps all of ACES goodness, no need to kneecap the dynamic range into display refered

Only means that in the future if i add the DCTL and then see apperant clipping, i will push a bit harder rather than assumeing.... using the 3x3 matrix and then playing is sure to reveal everything, takes longer tho, hence my leaning on the DCTL

and btw, i did say "seems to be clipped", as in i did not spend any time looking further, and did not say catagoricaly that is was clipped, only that is "seems tpo be clipped" when using the DCTL... you went a bit further than i did.. good for you to take the time... and i'm glad i added a qualifier to my thoughts...

i'd still thiink it could have been stopped down a bit as there's a ton of detail in the shadows already

 

[Marc Wielage]

Final image stayed Legal in the vectorscope for the Red channel.

That is right on the edge of not being legal. I would back off on chroma another couple of points. And I suggest you do not trust the internal scopes in Resolve or any other program. External scopes -- assuming they're calibrated -- will generally be a lot more accurate.

Note that this kind of thing is very, very typical for tons of footage out there. This is not just a Red problem -- it's just an issue of life in general, where the spikiness of certain kinds of florescent bulbs wreak havoc on cameras. They did it in the film days, too.

 

[rand thompson]

That is right on the edge of not being legal. I would back off on chroma another couple of points. And I suggest you do not trust the internal scopes in Resolve or any other program. External scopes -- assuming they're calibrated -- will generally be a lot more accurate.

Note that this kind of thing is very, very typical for tons of footage out there. This is not just a Red problem -- it's just an issue of life in general, where the spikiness of certain kinds of florescent bulbs wreak havoc on cameras. They did it in the film days, too.

Marc,

Thanks very much for what I know is Professional advice from a seasoned working professional. Unfortunately , I am only using A laptop screen( MSI Dominator Pro GT72S)to make these grades. Hopefully a professionally calibrated monitor with a very low Delta will be in my future. When I grade things on this forum I really don't know what you guys are actually seeing. I believe that most on the forum would have a properly calibrated monitor so I will take your word for any problematic grade that I may post.


I am still learning Davinci Resolve( Just got it to properly install on this laptop about 5 months ago) so I know that there is a lot that I still have to learn about it.

 

[rand thompson]

Marc,

Re-grade using 35-sat and new primary settings with more legal reds.

[/url]Capture by rand thompson, on Flickr[/IMG]

[/url]Capture 13j by rand thompson, on Flickr[/IMG]

[/url]Capture 8j by rand thompson, on Flickr[/IMG]




Capture 9j by rand thompson, on Flickr




Capture 7j by rand thompson, on Flickr

 

[Marc Wielage]

Marc, Re-grade using 35-sat and new primary settings with more legal reds...

Yes, that looks pretty good to me. One trick is that there's something like 11 different ways to adjust overall levels. I'm using the Highlight control a lot more than I used to, but for some things the Log Highlight control can help, for others the High Soft Clip can work, and sometimes you have to just bring the whole thing down overall and then bring all the mids and lowlights backup to compensate (what I call "backing in to a correction"). Every situation is different.

I'm sympathetic -- bright neon/florescent blubs are tough, particularly when they can't be on dimmers. I do often request that DPs watch the practicals on sets and have them on dimmers just so we can avoid having them go "nuclear" during the scene. Sometimes, what looks too dim to the naked eye looks fantastic on camera, particularly with a bare light bulb on a chain or a background light on a table. I would generally rather it be about 1/3 stop too dark than 1 stop over, because we can always overexpose it in post if you really want to slam it.

 

[Stacey Spears]

I would really like to shoot the Cars Land neon lights and see how they look in HDR. Its beautiful in person.

 

[rand thompson]

Yes, that looks pretty good to me. One trick is that there's something like 11 different ways to adjust overall levels. I'm using the Highlight control a lot more than I used to, but for some things the Log Highlight control can help, for others the High Soft Clip can work, and sometimes you have to just bring the whole thing down overall and then bring all the mids and lowlights backup to compensate (what I call "backing in to a correction"). Every situation is different.

I'm sympathetic -- bright neon/florescent blubs are tough, particularly when they can't be on dimmers. I do often request that DPs watch the practicals on sets and have them on dimmers just so we can avoid having them go "nuclear" during the scene. Sometimes, what looks too dim to the naked eye looks fantastic on camera, particularly with a bare light bulb on a chain or a background light on a table. I would generally rather it be about 1/3 stop too dark than 1 stop over, because we can always overexpose it in post if you really want to slam it.

Thanks Marc

I know what you mean, everytime I think I've found the most efficient tool in Davinci resolve to accomplish a specific task, I will find 4 or 5 more that are more efficient and take less time. I'm going to also start trying the highlight controls more now that you mentioned it. I find I use the Colorwheels "Primary Bars" and "Primary Wheels" color correctors when I need the "big honking" corrections and the Colorwheels "Log" color corrector when I need to "Fine tune" things especially for "shadow/lift" corrections.


The hardest thing to do in the neon sign grade above, was trying to make that yellow/orange area red. It was like trying to mix multiple paints to get that one shade of red.