| melaysho.com |
|
|
|
| DESCRIPTION My design was influenced principally by the projector's physical attributes. In this case access to the film gate was easier from what was originally the the lamp position. Because of this the image would need to be reversed either during capture or in post production. My first prototype used two brass plates, one for the sensor and one for the lens. Visualize if you will a micro view camera. In a bench setup this proved to be a working solution. However implementation proved impossible without having to discard half the projector (literally cutting it in half) and migrating the take-up reel. Trying to approach from the front (original projector lens position) with a lens/sensor assembly such as this was prevented because the assembly could not be brought to centre on the film gate. This led me to consider the tiny plastic lens that came with the webcam. It proved to offer a solution. The digital capture process I decided to use was a step and repeat one where the transport of the film (the projector) signaled the software/sensor when to make an exposure via a left mouse click trigger. Experimentation showed that the software/sensor took just under a second to record the image information, flush its buffer and be ready to record another image. I allowed 1 second. During this time the claw was completely disengaged from the film sprocket. This effectively determined the speed of the film transport. At this speed it took about another second for the film transport to move another film frame into position in the film gate. A 50 foot reel is about 3600 frames x 2 seconds = 7200 seconds / 60 = 120 minutes. Thus the word "fast" will never apply to this machine. It is slow but the capture process can be stopped and accurately resumed at will. In what follows is the solution I ended up building. I removed all parts of the projector that were superfluous (lens, motor, switch, light housing, electrical harnesses, etc). The stripped projector was then securely bolted to a wooden base that would also hold the adapted transport motor.
A brass lens/sensor plate was built to provide two axis of shift. The plate was held firm by 4 springs over the 4 bolts that together with the plate's loose fit bolt holes allowed a small amount of swing/tilt. This made centering and aligning the assembly with the film image in the gate easier.
The plastic webcam lens came mounted in a threaded tube that allowed for manual focusing and although it could focus down to about 30 cm this was not close enough for the purpose I had in mind. Essentially my solution was to use an extension tube. The search was to find a suitable diameter plastic tube that could be press-fitted between the sensor and lens. I found it in a cap for a dry eraser marker. The next step was tedious as in order to determine the correct distance between sensor, lens and film took trial and error which meant removing and remounting the assembly numerous times. Likewise press fitting the lens into the extension tube and insuring it was parallel to the sensor plane. I had no idea what the focal length of the plastic webcam lens was. Measuring the sensor diagonal and estimating the angle of view of the webcam lens gave me some figures to do some rough calculations which helped me get into the zone but getting on target took trial and error tweaking. The effort was compounded since there was no physical lens iris (f-stop) to increase depth of field.
CAPTURE SOFTWARE I used the ArcSoft capture software that came with the Nexxtech* webcam kit to setup the sensor/chipset. Importantly, the software's auto functions could be turned off.
The software GUI presented a main capture screen which included a button to capture a still image. Using the computer's normal mouse the cursor was positioned over this button. When the machine runs it sends a left mouse click signal to the computer through its own (PS2) connection to the computer. The computer does not discriminate which mouse is doing the clicking and thus the cursor being positioned over the button activates the capture software routine - records the frame image and stores the resulting Bitmap file sequentially numbered.
* Nexxtech appears to be a house brand of The Source electronics stores in Canada.
POST PRODUCTION I used Adobe's Photoshop, Premiere and After Effects for processing the still images and rebuilding the movie digitally. To capture a Super8 sound track I used an older Bell & Howell Filmosonic projector which had analog audio output. The camera used in the sound examples unfortunately was not blimped and sound was recorded using automatic gain. I used Sound Booth to clean up the sound as best I could. I processed, cropped, etc., in Photoshop preparing the still images for import as sequences into Premiere. In Premiere I built the digital version of the film original and in the case of Super8 sound films the sound track was synchronized. While editing I discovered something I had not anticipated. While each frame is positioned in the gate and the claw disengaged during image capture the projector's film transport does not stop. The film continues to be transported. The film loops together with the film gate's pressure plate should prevent movement of the film during its short stop at the film gate. I observed this to be so but intermittently the film would move ever so little (<=1mm) that the result at real time speed induced a kind of "shake/flutter". I used After Effects to remove this effect. Not to be mistaken for the original camera shake which is all mine. SAMPLE RESULTS Note that the sample clip below has been reduced in size and compressed for the web. The embedded version plays mp4 version, the link opens a window to play QuickTime (.MOV) version.
|
|
|
|
||
|
Requires HTML5 capable browsers. Mouse over image area Otherwise: Apple QuickTime version (11.93MB MOV) |
| Copyright © 2016 Chris Milejszo All Rights Reserved |
