CG in Video II
I've come a long when when it comes to CG in video. I finally have managed to get soft shadows and graphics that fit into the frame better.
It may not be noticeable in the frame but that fuzzy white at the bottom of those columns makes the soft shadows that lay over the asphalt. It helps sell the effect that I'm standing next to an assortment of columns which should look fairly familiar by the way.
If you notice the shadows my feet make you can get an idea for the lighting. The Hemi light works well for this cloudy day. I'm going to have to go looking for cloudy days to film on because I can't emulate other types of outdoor light yet. ... yet.
CG In Video
The problem of adding a Computer Generated element to video is often an exercise in guess and check. This should not be so. The problem is well understood. Imagine a real object and replicate the result of filming that object.
If we explore this from a scientific stand point and grab our physics books I do believe realistic CG composites are possible by going down a check list.
The CG in the video must also take on the Noise of the Video.
If it doesn't the CG elements stands out far to crisply from the surrounding footage. The colorspace of the CG element must match the host host format as well. The video footage can be studied itself. Luminance channels,interlacing, colorspace maps and gamma values all yield information that should be useful for analysis and insertion of CG Objects.
Unfortunately, I don't yet understand how. This process will take some considerable investigation.
Beyond that there are many things to measure on set before even thinking about recreating the illusion of an object without the guess and check.
- Shadow Test (with stand in object)
- Ambient Light level
- Temperate of the Light
- Intensity of light at object and at camera
- Perspective of proxy objects must be measured
- Note number and location of Light sources
- Camera Location
- Camera Angle
- Focal length
- Lens
Though some physics research about materials and their optical properties it may be possible to generate a best guess as to how to model and render the object with the video. The other question will be how to translate that data into values inside of blender.
Writing some sort of pyScript helper might be a good aide, but still the scope of the study is quite broad.
I'll contemplate finding a white cube to place in a scene for shadow and light measurement. It should help reflect the radiosity going on if that is at all pronounced.
It will be a challenge but it's worth the study.
When filming a shot for CG insertion we can help our selfs out later by getting footage with markers in place and measuring those markers.
I'd like to emulate cloudy and a sunny days.
Noise must be mimicked it's that same noise that makes a single frame stand out. You can tell when we freeze the frame. The noise doesn't look right.
Theoretrix
Exploring the edge of calculation and the beginning of philosophy. Walk along the shores of knowledge where ideas meet dreams. Travel to the limits of imagination and then a bit further. Welcome to Theoretrix the show where we explore the hypothetical, the edge of the possible and imagine the beyond.
Notes in program from.
why not write the program as I learn.
If material is explicit and rules can be written it would be a waste not to.
End of year stats movie
If your business can't survive without keeping secrets...
I think you had better rethink your business.
I know it can be argued that for machines to pull off these kinds of things would require some tremendous improvements in functionality. One can argue they're inefficient, difficulties and organizational issues. And all the work that would go into making it all work.
...
But I can also state that this human based system of functioning is complicated, difficult, inefficient and hard to mobilize to meet broad goals.
The difference?
One system I can make progress on....
.... the other I can not.
I did finally get CG into video today.
I mean in a way that works.