[jfrancis]

Suppose I have a photo of a cardboard box that I like (with labels, and nicely lit) and I want to insert it realistically into a scene and have a nearby maxwell mirror sphere reflect it.

If I map the photo of the box onto a virtual box it will be doubly-lit, since the box is already a photo of a lit box, and the virtual box will be further lit by the virtual scene.

I wonder if I light the scene with a white Maxwell box, then use the new Photoshop DIVIDE blend mode to divide the photo by the white box lighting, causing the photo of the box to be extra bright...

...if I then map that extra bright box photo onto a virtual box if it will look fairly normal in the scene.

Or is there an easier way?

[RichG]

I've had the same problem. I ended up mapping the photo as an emitter and adjusting the brightness until it's not blowing out. You can always comp the actual photo back in afterwards for true fidelity to the original.

[jfrancis]

I was wondering about that approach. Haven't tried it yet. I'm glad it seems to work out for you. I'll check it out.

I guess there is no really correct thing to do since the photo contains specular highlights and mapping it or using it as an emitter treats the specular highlights as diffusely painted on, but four our purposes it's probably good enough.

[RichG]

Yes, it's always going to be a kludge but the advantage over purely comping it in post is that it does contribute a little in terms of a very weak emitter giving the effect of diffuse bounce from the photo to the surroundings.

[tom]

Sorry, I'm not sure I understand something from your descriptions. Is there a reference photo of this thing?

[jfrancis]

A invisible CG figure casts shadows onto the board. The figure matches a real figure in a photo.

So far so good.

Here comes the question:



If I want the photo to reflect in the knife blades I need a virtual figure.

I want to map the virtual figure with the posed real photo of the figure.

Should I make the mapped virtual figure a weak emitter? Or should I make the photo extra bright (maybe by dividing it by a diffusely lit white figure) so that the extra bright map looks 'normal' when the virtual light shades the figure?

[jfrancis]

In the case above I chose blurry knife blade 'mirrors' and a very approximate painting to stand in for the photo, but lets say I wanted better mirrors and better mapping (I know one map would not work from the mirror's POV, but set that aside)

[jfrancis]

Another version of the question:

Let's say you have camera projected this simple scene onto simple forms.

How do you put a mirror ball in the scene that reflects the floor below and the box on the side?

The photo is already lit. If you map it onto simple geometry and light it the image will be doubly-lit.

Do you make it a weak emitter?

[tom]

Do you make it a weak emitter?

Yes, that's the correct answer. You should avoid re-illuminating the real-life baked texture. How much weak? No problem. You should build HDR textures from different exposure photos so, their intensity and range will be coherent when used as emitters.