First of all when shooting set your camera to have both exposures that you need, in my case, interior and exterior, properly set.

If you choose to shoot it in FullHD you will be locked to the 30fps speed, I prefere shooting in HD720 with 60fps.

As I show in the beginning of the video, Magic Lantern output a blinking video, one frame will be underexposed and the other will be overexposed. Having that in mind if you shoot in 30fps you will have 15fps of each exposure footage, and less information to bring back to the original frame rate. If you shoot in 60fps you will have 30fps for each, after the HDR merge process but you will have a finished 720p with clean frames as output playing in real time in 30 fps, instead of a fullHD 15fps that you will need to interpolate with some third-party plugin like Twixtor for AE or Chronos for Nuke.

1. Make a new comp with 720p 60fps footage.

2. Duplicate the layer.

3. Get rid of the first frame of the duplicated layer.

4. Select both layers and right-click, Time, Time Stretch.

5. Type 50 as the Stretch Factor.

6. Now you will have splited the over and the underexposed frames in different layes.

7. Duplicate the Underexposed layer and name it as Mask.

8. In the Mask layer add a Levels effect and adjust it in order to have the biggest contrast between the bright areas and the dark areas. Turn on black and white clipping with you are using 32bit comp as I did.

9. Add a Fast Blur effect and rise a lot it to have a nice transition, in my case I used 60. Check Repeat Edge Pixels.

10. Assign the Mask layer as a Track Matte for the underexposed layer. Render! The video will be played at 200% speed because your comp is set to the playback speed of 60fps.

11. Render it out as an image sequence and re-import as a 30fps footage to have ir in real time speed!

12. To get the sound of the original footage, just drop in to the new 30fps comp the original footage and hide it, leaving just the sound active.

This workflow worked great for me, but you might need to tweak some settings to match your footage and lighting conditions.

Make sure you have a clean mask to work with. I tried other ways but I think this was the most easily and effective.

Below you can watch the final result!

Hope you enjoyed! Any doubts let me know!

I used a Canon 600D (T3i) on 720p 60fps video mode. Magic Lantern does shoot one overexposed frame and on underexposed frame, giving an output of a blinking video.

To make it work you will need a post processing software such as After Effects or Nuke to extract every odd and even frames and blend them together with a luma key applied as a mask to key out the overbloom areas and have there the information in the underexposed sequence.

Below I show you the both frames and the final result.

I made the talking smile potato, the iPad on the face of the guy, the painting mouth talking and the end logo and title animations.

I did the color correction for the movie, the design for the titles and the animation.

The idea for the project is from Marcel Maineri and Julie Bazacas and It was directed by Guido Antonini.

To follow this project access the page http://projetopare.tumblr.com/

I did the color correction for the movie and the text information.

The project was filmed with the pink background so I added a little bit of pink tone to the skin color of the women to bring them more into the “pink world”.

Below you can watch the movie from the director’s vimeo page:

This month I’v done a job for the videoclip of the Brazilian band “Robo Gigante”. Below I’ll be showing some of the effects that I made for the project with some before and after images.

The replacement of original images that were on every TV on the videoclip with images of the band:

In the scene below added more cars in the traffic:

In this scene I did color correction and a digital interface to simulate the vision of the robot:

In addition of doing the color correction in the whole movie:

Below you can watch the entire videoclip, produced by Zeppelin Filmes and animated, rendered and composited by Dr.Smith!

The tracking was challenging to do because of the low light footage and the non-linear camera movement, I used Autodesk Matchmover at first and I got a very solid solution, but the scene scale wasn’t matching that well. So I tried again using PFTrack and I got it very nicely even handling the lens distortion that the Sony EX1 gave me because I used the widest focal length.

To match the light conditions on the scene I made two 360º HDRI images for the CG reflections, one with lights on and another one with lights off.