[Bf-compositor] Clamping (Mix, Math, RGB Curves)

Diego Gangl dnicolas at gmail.com
Sat May 24 00:07:28 CEST 2014


Hi all, I don't know how Blender's handles colors internally but
wouldn't this work?
http://www.pasteall.org/pic/show.php?id=71736
(or are the V values already mapped to 0-1 from the separate node?)

Cheers,

2014-05-23 18:46 GMT-03:00 Sean Kennedy <mack_dadd2 at hotmail.com>:
> Could we set the value of the clamp? It could come in as defaults of 1.0 and
> 0.0, but we could alter it to whatever we wanted. At R&H, most of the time
> we clamped brights around 14-16. We'd only turn it down under that if we got
> bad pixels when rendering.
>
> If you had a superbright red, like R 12, G 3, B 4, and you put a clamp of 1
> on it, obviously it's going to set each value to 1, so what was red now
> reads as white. But it would still be useful.
>
> After Effects clamp tool
> http://cdn.aescripts.com/media/catalog/product/cache/1/image/800x600/040ec09b1e35df139433887a97daa66f/f/t/ft-clamp-ui_1.png
>
> Nuke clamp is in the color grade node
> http://docs.thefoundry.co.uk/nuke/63/pythondevguide/_images/getMinMax_01.png
>
> I'm not posting the images to say "this is how we should do it", but rather,
> to say it is useful and professional compositors would indeed use it often.
>
> In fact, I was working on a shot this week that could have used it. I was
> given stock footage from a client, and I have no idea how much they'll color
> correct it in post once I deliver the final vfx shot. SO I have to make sure
> all my blacks are in line with what is in the plate. I put a gamma on the
> plate just to view the black levels and see how they match up when pushed
> really far.
> https://dl.dropboxusercontent.com/u/10553822/clamp_blacks_01.jpg
>
> Since we don't have a clamp, I ended up putting a mix node right after the
> stock footage node, setting it to Lighten, and making the color over it
> black. then I slowly raised it while watching until the blacks all came in
> line. Would have been much easier to sample the value of the darkest dark I
> wanted in the shot, throw a clamp node on with the black clamp set to those
> values.
>
> In fact, at R&H, the Black Clamp node let you set an individual value for R,
> G, and B. And it had a color picker so you could just sample the image and
> the values would fill in automatically.
>
> Anyway, just throwing out my $0.02. Obviously I've found a workaround and
> can live with it, but it would sure be great to someday see a clamp node.
>
> ________________________________
> Date: Fri, 23 May 2014 14:21:40 -0700
>
> From: troy.sobotka at gmail.com
> To: bf-compositor at blender.org
> Subject: Re: [Bf-compositor] Clamping (Mix, Math, RGB Curves)
>
>
> On May 23, 2014 1:56 PM, "Bartek Skorupa (priv)"
> <bartekskorupa at bartekskorupa.com> wrote:
>
>> When I say: "clamp blacks" I always mean: set every value below 0.0 to
>> 0.0.
>> When I say: "clamp whites" I always mean: set every value above 1.0 to
>> 1.0.
>> Every compositing artist will know what I mean.
>
> If we can generate a clamp / scale node, I am all for it.
>
> I am saying there is no concept of white in a scene referred model, and as
> such, you have to be careful as to not write nodes that break the system.
>
> It is that simple. No white. No brighter than white. So sorry, “Every
> artist...” is actually “Every artist unfamiliar with HDR work.”
>
> So a clamp node _must_ not make the erroneous assumption that 1.0 means
> anything special, and rather provide for an arbitrary value range.
>
> If this all seems like speaking an alien language, it is precisely why I
> chimed in and said “Whoa.”
>
> There are already nodes that are broken because of erroneous assumptions. It
> would be a shame to make more.
>
> With respect,
> TJS
>
>
> _______________________________________________ Bf-compositor mailing list
> Bf-compositor at blender.org
> http://lists.blender.org/mailman/listinfo/bf-compositor
>
> _______________________________________________
> Bf-compositor mailing list
> Bf-compositor at blender.org
> http://lists.blender.org/mailman/listinfo/bf-compositor
>


More information about the Bf-compositor mailing list