Wolverine-Hawkeye Telecine

[David Brown]

Thanks Stan for taking time to check that. I will be testing it.

BTW when I asked if some of the values are not decimal, the white balance seemed odd. This is an excerpt from an iccf file, saved from IC2.5, shown with the IC2.5 display.

I've not had problems with white balance in either app. I parsed through the iccf and everything matched EXCEPT white balance.



I'd be careful to edit those numbers without a "conversion"? Not to worry, I tend not to edit these files. Who knows what an extra space or lost > would do!

[Stan Jelavic]

David, the values in the iccf file are 0 - 255 8 bit colors. The IC displays decimal values 0 - 4.0. For example:
red = 2.33 = (2.33/4.0)*255 = 213
I assume they want the iccf file to have true 8 bit RGB values and the display can be whatever, 0-4, 0 - 10. If you ask me I would put the rgb values right on the display but I guess when you go to 10 or 12 bit that could be an issue, the ergo would have to change. This way no ergo change is required.

[Kamel Ikhlef]

Hello everyone, just a hello.
I hope that you all are ok. I left out the digitization of my films a bit.
But I will resume soon. I think in a month.
I am happy to read you and follow your various experiences (DSLR). Still very interesting.

[Stan Jelavic]

Glad you are back Kamel.

[David Brown]

Hey Kamel!

Stan, after hours of testing......HDR is OK. Nice touch adding the decimal increments for the Low setting.

Also it will now go negative so it can be the bright value, if I choose.

I had a list of changes I thought would be nice for the app, but now I think they would only be useful when testing for hours!


I found that the auto exposure box can get checked without my consent. I can't replicate that but it seems to happen when stopping, changing settings and restarting without exiting. Watching the capture display, it is seen within the first 2 or 3 images and the display no longer flashes the two exposures. Then, just exit and restart.

What does the 64bit button do? It will cause an error if selected when running in 32bit. Just exit and restart, or don't push the button!


Otherwise, I'm back to capturing!

[Stan Jelavic]

The app turns off auto exposure when setting the low exposure to prevent auto exposure from compensating. Then it turns it back on for high exposure. So if you stop and check the settings, auto exposure could be off or on depending on which point you stopped. The 64bit saves the images in 64 bit mode. It has to be done up front. It will give you an error in any running mode. You have to stop before changing to 64bit.

[David Brown]

Just to be sure I understand, the first exposure is with shutter on automatic and the second is manually offset by the Low setting?

I have been overexposing the first and underexposing the second (via Low), with the shutter.

The odd thing is when I find the auto exposure on, both images have been exposed and saved exactly the same.

If I use 64 bit, then I also need to select the bit64 button as well?

I now plan to have all the information up front and not interrupt the app and change anything.

[Stan Jelavic]

Yes the first exposure is in auto but can be controlled by
Auto Reference
and
Auto Max Value

I set these so that the image is overexposed.
The second frame will use whatever the last total exposure was and will subtract the increment.
And then will switch back to auto mode. I think this is a good way to have the auto exposure working and controlling the average of the two settings. But we can change to to whatever we want as long as we stay within the api bounds.

Good question about the 64bit button. I essentially force image write to 64 bit but not sure if that will override the initial setting. Can check.

[David Brown]

I'm still testing this. The overexposure range was not adequate for my test without adding gain. I went the opposite way and set the 1st frame as underexposed, and the 2nd frame as overexposed. That's a negative value. More to test.

The 64bit mode is working as the Enfuse processing was only ~1 second per frame, now it 45 seconds! At that speed I can read the errors, or in this case warnings. They are probably not important, just estimates for settings. Here's the capture.

[Stan Jelavic]

Looks like you are getting there "slowly".

[Bill Putt]

got my hawkeye pretty much assembled and tested. now working on tweaking. biggest issue right now is glare from the part of the led light source which is not covered by film(see pic 1). Any way to correct for this? I played with config settings in ic capture with no luck. changing focus of lens moves where the glare falls, but it's still there. perhaps putting down tape where it's glaring from? just means i'd have to change between 8mm and super 8 batches.


also wanted to report that my model of wolverine seems to be a bit different as the L shaped USB cable goes right into the gear, so i had to get a different cable

and i can't mount the switches where the og screen was cause the handle of the machine is in the way. I could Dremel out the plastic. Haven't decided yet.

[Stan Jelavic]

Hi Bill, my records indicate that you have the Hawkeye kit with the 16mm lens. What camera did you get and what spacers did you end up using for the camera? From the picture that you posted the camera seems to be out of focus which will make the lens flare (lens glow) worse. Even when fully in focus you can get the flare under some circumstances. There is a gate mask that you can use that reduces that significantly. Do you have access to a 3D printer to print out the mask. If not I can get some printed from an on-line shop. Can you try using the standoffs for the switches?

[Bill Putt]

Stan,
I have a DFM 72BUC02-ML. Here are pics of how I have it mounted. This is my best interpretation of the instructions. I have a friend with a 3d printer who could make some up. Are they on thingaverse? I don't see a gate mask in your files list on there. Keep in mind the film in the picture is a black and white negative cause it was recently developed. looks pretty clear to me. I'll figure out a way to mount the switches. mostly just letting you know the the issues I ran into. I might just mount the front part on a completely homemade back part.

[Stan Jelavic]

Ok that all looks good Bill. Never realized the film was a negative. I inverted the colors and now I can see that the focus is pretty close. Lots of noise in the image. Here is the mask files:
https://drive.google.com/drive/folde...UU?usp=sharing

[Chris LaRocque]

Hello Everyone

It's been a while since I was last active here and I'm happy share the first revision of my autocrop tool for STD8 stock. The script works when the frame includes the perforations on the left and also the space to the right of the film stock. Normal orientation is required so there is a rotate flag, files must be 1920 x 1080, and you should specify absolute file paths for all inputs. The two avi files are an example of raw and cropped film, converted to 16 fps video without dark frame, scaling, or border.

I am querying The Imaging Source about artifacts found in dark frames and images so hold off on processing dark frames for the moment. I'll post more information soon.

Application flow:

Frames are verified to exist.

Optional Output path is verified to exist.

Optional Dark frame path is verified to exist. All files found at this path are presumed to be dark frames, must be oriented correctly, and are not effected by the rotate flag. Dark frames are averaged and the result subtracted from all input frames.

The right edge is located and the value "pixels per mm" is calculated based on the std 8 stock width of 7.975 mm.

The Perforations are located and the center frame is calculated based on centerline Y being midpoint of the perf separation and centerline X being 2.47 mm right of the perforations right edge.

Frame crop is calculated as 4.5 x 3.3 mm with a 1.1 scaler applied to capture as much of the available image as possible.

The Cropped frame is optionally white balanced using average lumanence (LAB format), rescaled proportional to the height of the input image (1080), and then bordered with black rectangles to the right and left sides, adjusting it the original aspect ratio.


Example relative path command line tested on Linux with bash. Use absolute paths preferentially because I haven't exhaustively tested relative paths.

ls Raw/*.tiff | AutoCrop_v1.0.py -v -o Cropped

Output of AutoCrop_v1.0.py -h

usage: AutoCrop_v1.0.py [-h] [-f FILE [FILE ...]] [-r] [-o [OUT_PATH]] [-k [DARKFRAME]] [-d] [-s] [-w] [-b] [-v]
[-c CHECK_FRAMES [CHECK_FRAMES ...]]

Crop image files automatically.

optional arguments:
-h, --help show this help message and exit
-f FILE [FILE ...], --files FILE [FILE ...]
Files to read, if empty, stdin is used
-r, --rotate Rotate images by 180 degrees.
-o [OUT_PATH], --out_path [OUT_PATH]
Output File Path. location of input file is used by default.
-k [DARKFRAME], --darkframe [DARKFRAME]
Path to folder containing Darkframes that will be subtracted from each image
-d, --debug Enable display of each stage of image isolation
-s, --scale Scale cropped image vertically to height of original image
-w, --white_balance White balance based on average luminance of cropped image.
-b, --add_border Add border to width of original image
-v, --verbose Annotate pipeline stages; list each file name and path as it is processed.
-c CHECK_FRAMES [CHECK_FRAMES ...], --check_frames CHECK_FRAMES [CHECK_FRAMES ...]
A sequence of frame numbers. Sets debug only for those frames to allow examination of the pipeline
process


To do:

Add Frame orientation check and correction.
Add S8 processing
Evaluate / Implement alternate frame keying strategy.
Improve vertical frame bounce control. Y axis centerline changes slightly when locating the perforation outr dimensions.
Test code on windows.
Add TCam-Capture camera interface to pull images from Wolverine directly.

I'd appreciate feedback. If anyone wants to help I'll set up a repo on Github.

Regards

Chris LaRocque