Matching Levels and Colors from Different Sources

requires Vegas version 4

One of the more difficult tasks in video editing is trying to match footage from the same event shot with different cameras. Not only will the different cameras likely process the colors somewhat differently because the cameras are in different positions and the distance from the subject isn't the same, likely you'll also be dealing with different lighting levels. Worse if you do a white level adjustment in one camera and forget to do it in the other, results can be disappointing for sure. Special thanks to Dave Kaffine and G Randy Brown for allowing me to use their videos as source material in this and the previous tutorial.

Images one and two are the same scene shot from two different cameras. So how do we go about bringing them closer together? Its a multiple step process and for sure trial and error is involved and also some judgment calls trying to decide which is truer to what the subjects actually looked like. I must admit this project gave me some trouble and while I think I made fairly decent corrections to illustrate the concepts, I for sure could have spent more time getting them closer still.

As as first step, see what the two shots have in common. In this example all that's the same are the subjects! The walls, the stage floor, even the pianist's shirt are very different, making it hard to have a starting point to adjust off of.

Thankfully, Dave sent me some similar shots so I could get a better idea of the wall and floor coloring. Absent that you just have to use common sense and look for telltale objects like the sheet music on the top of the piano which should be close to white and the piano itself which mostly likely is black or maybe a very dark mahogany color. The walls are more of a guess, and can and probably were influenced by stage lighting. Well I like a challenge, so lets see what we can do. The first step is drop both images on a video track side by side or on separate tracks if you doing videos and not still images. Next decide which best represents reality.

The scopes feature in version four can help. As you see above the Histograms are different in the midrange and highlight areas, the middle and right portion of the the graphs respectively, while the shadows or left third of the graph are fairly close. So next we'll use color curves to attempt to bring the Histogram of image two closer to image one's Histogram. As a initial step I just moved the center midpoint a little to the left which will shift some high range pixels closer to the midrange as they are in the Histogram of image one. Also note I pulled the top right end of the curve down slightly and the bottom left has a slight jog. That helped some, but we got a long way to go.

To remove the severe orange cast, we'll try first to use the Complementary Color eyedropper starting with the Low colors wheel. I tried clicking on several places on the piano and wasn't that happy with any result, in fact there was hardly any change at all so I just manually pushed the it down and very slightly to the right toward blue and cyan, shown as 'B' and 'Cy' on the wheel.

Next I Tried the Complementary Color eyedropper on the sheet music to adjust highlights, didn't shift the image as much as I hoped. Worse, no matter what setting I tried for any of the color wheels, the man's shirt remained too gray instead of getting it more blue like I wanted. So to help things along I just eye balled it and ended up pushing the High wheel fairly far towards blue. Better, but I had to avoid going too far or the entire image shifted from a orange cast to a blue cast substituting one problem for another.

Color Corrector Secondary filter is another new feature where you can select to adjust just a narrow color range. I used the Complementary Color eyedropper, lower left, right under the color wheel then dragged it to the preview window to draw a outline of part of the man's shirt sleeve. After several attempts I was able to get the adjustment I wanted by also changing the Rotate hue setting along with the Saturation, Gamma, and Gain while also going back and forth nudging the same settings in the primary Color Corrector. Starting to make progress with the orange cast now gone and the walls and stage floor close to the same tones as image one.

Next I dropped a new Color Curve filter unto the timeline and selected only the blue channel instead of the default RGB. As the name suggests selecting just the blue channel option won't alter the Red and Green channels as it does when all three channel are present like they are in the RGB settings. After 30 seconds or so of slowly dragging the curve set points around on the curve I settled on the curve you see.

Above is the before and after shots of the images and final Histograms. Much closer than the two originals, but not exact matches by any means. The problem was if I got the shirt bluer the walls would go further out to tone. So while dragging the curve around I focused on the skin tones in the man's face. I also 'cheated' a little and threw in a couple of older filters. Levels changing Input start:44 Input end: 960 Output Start: 62 Gamma: 1055 and Color balance: R:110 G:25 B: 450,Dave sent me a nice shot of the girl's dress, however attempt to bring out more green only threw the man's shirt further away. Further adjustment could alter image one slightly to bring it closer to images two. Again a trade off. You could also attempt to further refine to peaks in the midtone region