Announcement

Collapse
No announcement yet.

Wolverine-Hawkeye Telecine

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Hi Bruce, I do agree that if Wolverine did not have artifacts it would be a quite nice and affordable scanner. The issue is that there is absolutely no way of fixing that. Possibly smaller screen and some post reduces the effects so that may be good enough for many but not for me. Even on my 17" laptop.
    Looking at the 2 pics that you posted I do not see much color difference. Maybe it is my old eyes. The wolverine image looks sharper than the UX178 (that is the camera used here with Hawkeye).
    But if I put that frame under the microscope I see about the same fuzziness as the UX178.
    Click image for larger version  Name:	ux178-teslong.png Views:	0 Size:	1.51 MB ID:	24676

    The Wolverine image is bad in my opinion. Look at the reflection from the horses. It does not look natural.
    Now if I crank up microscope sharpness to 50 this is what I get.
    Click image for larger version  Name:	wolverine-teslong.png Views:	0 Size:	391.5 KB ID:	24677

    The images are very similar in artifacts.

    As far as the exposure goes Wolverine has a big problem with the auto exposure hunting. I get it at least once per clip. Here is an example.
    https://photos.app.goo.gl/mL8th6yCnMWqyBYp9

    Additionally, the exposure is not consistent as I mentioned before. Here are examples.
    The hills and the sky are overexposed here.

    Click image for larger version  Name:	exposure1.png Views:	0 Size:	1.20 MB ID:	24679

    Here, my cousin's face is under exposed (too dark).
    Click image for larger version  Name:	exposure2.png Views:	0 Size:	1.22 MB ID:	24680
    But the hills behind are visible and barely visible on the Hawkeye shot.
    Not saying that Hawkeye is much better but it does have an HDR option and I still maintain that the exposure with Hawkeye is more consistent and stable. I actually made the exposure a bit on the dark side by setting the reference to around 50. It is a compromise.

    Shifted the Wolverine colors away from yellow.
    https://drive.google.com/file/d/1-AF...ew?usp=sharing

    Here is the Video Fred change:

    #COLOR AND LEVELS PARAMATERS
    #----------------------------------------------------------------------------------------------------------------------------
    saturation=1.2 #for all outputs

    gamma= 1.2 # for all outputs

    blue= 10 red= -5 #manual color adjustment, when returning result3 or result4. Values can be positive or negative







    Last edited by Stan Jelavic; January 01, 2021, 04:57 PM.

    Comment


    • Good point Bruce!

      Anytime Wolverine captures can be seen smaller, they look better! I watched this side by side on a laptop using the google viewer. Wolverine wins hands down. Now, downloaded and on my desktop, it loses.

      That's why I bought the Wolverine, I had just watched a you tube sample in a small viewer. Lucky I did, there is nothing close to it's price on the market. Most of that luck came after finding this forum with Stan's mod, the Hawkeye.

      Comment


      • Hi Stan and David, perhaps I commented badly, did not mean to make it sound as though I was criticising the Hawkeye. I do not disagree with your opinions, the huge plus with the Hawkeye is being able to start with very pure virgin high quality images (not sure of terminology) and being able to post process these to produce some very good end results (HDR is an option, remove blemishes, scratches, adjust colour & exposure, set desired sharpness, remove jitter, change resolution, crop etc.).

        In the end it depends on the importance of the film to the individual, how much time and money they wish to spend.

        Previously I Was not sure which camera Stan used for the Hawkeye clip, the colour is definitely more natural. The results of Stan's HDR test clips, that I have seen, using the UX178 camera, are very good, I also loved the end result colours.

        All the best
        Regards - Bruce

        Comment


        • No problem Bruce. Your comments prompted me to dig a bit deeper into the differences which is good. When I saw the image that you posted I was getting concerned that perhaps the UX178 was out of focus or I pushed the lens too hard. But looks like it is just the original resolution of the film. The latest wolverine clip with the color tweak looks pretty decent I think. Regards.

          Here are the images from the UX226 camera with no mask David:
          https://drive.google.com/file/d/1Vdf...ew?usp=sharing
          https://drive.google.com/file/d/1F1H...ew?usp=sharing
          Last edited by Stan Jelavic; January 01, 2021, 08:52 PM.

          Comment


          • Hi Stan and all.

            Stan thank you for the side by side video (it's not available anymore on Google Drive by the way, but I previously downloaded it).
            Sorry, I know I might be repeating observations already made in this long thread, but maybe an updated summary from time to time can be useful for next newcomers, too .

            If you examine single frames in your video the shortcomings of the Wolverine are clear: too much contrast and compression, as you already well know.
            But when watching the movie the Wolverine result is not bad at all. There are some terrible artifacts on the striped shirt of the girl but I suspect they have been introduced by post processing, can you confirm?

            Anyway, I still cannot focus why, but the Wolverine video looks amateur and "digital", while the Hawkeye one looks professional. Even though they both look good. When watching the video on my 46" TV the difference looks even a bit more accentuated, first because Wolverine artifacts are slightly more visible, but IMHO also because the "pro" feel of the Hawkeye video is more suited for the TV screen.

            I think I've seen the auto exposure hunting somewhere in my Wolverine, but usually I don't get it, and maybe converting at total zoom out including sprocket holes helps with this.

            Comment


            • Thanks Marco. Good comments. I removed the first side-by-side video but have a reduced resolution one still there. It just chews up so much of my google drive storage since I am using the Lagarith lossless compression instead of MP4.
              The striped t-shirt artifacts are present in the original Wolverine mp4 also.
              Click image for larger version

Name:	striped-shirt.png
Views:	414
Size:	1,009.7 KB
ID:	24730
              I think this stuff happens when sharpening is set too high. See the test frame I posted earlier. It has similar type of artifacts then there are alternating dark and light lines.









              Comment


              • Hi Stan,
                I think we are not talking about the same artifacts on the t-shirt stripes.
                Here are the ones I'm talking about:

                Click image for larger version

Name:	stripes.jpg
Views:	411
Size:	136.8 KB
ID:	24765

                It looks like some areas of the t-shirt are being moved around, to me this seems a side effect of some auto-stabilization algorithm or similar.

                Anyway, today I finally removed the small gear and cam, I confirm there was no retaining clip, but the small gear is very tight and I had to pry it with a screwdriver:

                Click image for larger version

Name:	cam_open.jpg
Views:	400
Size:	68.3 KB
ID:	24766

                Anyway everything seems to be ok there as far as I could see, according to your suggestions and everything I could check. I think the small friction I feel when moving the wheel is coming from the small spring itself when the cam moves up and down, but I cannot do anything about it and anyway as previously explained I think this is not the cause for my jitter.

                I concentrated again on the large gear and noticed another thing instead: I think it's not perfectly flat.
                I took a slow motion video of the turning gear, if you look carefully at its alignment with respect to the assembly cover at the bottom you can see that it is slightly oscillating:
                https://photos.app.goo.gl/PHMKRmdXNQBCNDvDA
                I captured two frames of the slow motion video to show the different position of the gear in two opposite moments of the gear period:

                Click image for larger version

Name:	gear.jpg
Views:	400
Size:	82.3 KB
ID:	24767

                I tried to push it flat as far as I could, with no success. Maybe actually the motor shaft is not perfectly straight?
                Anyway, now, I think it is possible that this slight oscillation causes a different pressure on the small gear and on the cam, in different moments of the large gear rotational period, causing the periodic jitter.
                I'm still not sure if I can find a solution... suggestions are welcome.

                Comment


                • Hi Marco. The stripes issue is present in the original but it is not as bad. The post makes the issue much more noticeable. Some tweaking of the script may be needed to reduce the effects. I have not spent much time with the script.
                  Getting back to your jitter issue you can try reinstalling the gears and the cam without the claw mechanism and then slowly rotate to check for any binding. Binding may cause the stepper to stall (exceeding the max torque) and cause it to hesitate momentarily resulting in jitter. The crooked gear may be catching on something (maybe the cam) causing the jump.

                  Comment



                  • Marco & Stan, could the erratic claw movement be coming from the cam? It looks like plastic and might be worn or flawed in at least 1 axis.

                    I'm referring back to October '20 and Stan's post #1118.

                    The distortion in the girl's t-shirt could have come from the dirt and dust filter that is in the post script. I don't know how to turn it off, but the settings can be reduced. That's a guess. I don't remember this clip but the only motion would have been from the hand held camera and it's hard to believe there was enough motion to cause it.

                    Stan, I asked for an unmasked and cropped image to see how much more magnification you have. You have abt 10% more (did not measure it) that's the advantage of swapping mounts for S8 and R8. I could have had a little more than I do, but I left some margin above the the S8's image to catch any framing jumps that happen. I'm using a V12 board and have not modified the camera mounting holes or the camera. To get closer, I would have needed to center the camera over the frame ( and build a second mount for R8). I'm happy with the captures from both S8 & R8.

                    https://drive.google.com/file/d/17Gm...ew?usp=sharing

                    Comment


                    • Hi David, yes it could be the claw although the noise that Marco originally posted sounded more like gear teeth catching on something.
                      http://mdv.eu/temp/MovieMaker_Problem_Short.mp4
                      Also the fact that is roughly happens once in 4 frames indicates the large gear and that gear looks warped. Possibly a few teeth catching on some part of the cam and stalling the stepper, recovering and another tooth catching on it etc. As the gear moves the clearance increases and the gear moves free.
                      Just a thought... really bizarre problem.
                      Your R8, S8 solution looks very good.
                      I also tried the hi rez with the BUC 02 and 16mm lens.
                      The bottom spacers are 8mm, same as the UX178 and UX226. Then V1 or V2 adapter board. Then 10mm spacers and then the camera.
                      Here are the captures.
                      https://drive.google.com/drive/folde...ip?usp=sharing
                      he cropped R8 is around 1600x1200 and S8 is around 1900x1400






                      Comment


                      • Originally posted by Stan Jelavic View Post
                        Hi David, yes it could be the claw although the noise that Marco originally posted sounded more like gear teeth catching on something.
                        http://mdv.eu/temp/MovieMaker_Problem_Short.mp4
                        Also the fact that is roughly happens once in 4 frames indicates the large gear and that gear looks warped. Possibly a few teeth catching on some part of the cam and stalling the stepper, recovering and another tooth catching on it etc. As the gear moves the clearance increases and the gear moves free.
                        Just a thought... really bizarre problem.
                        Thank you Stan, just let me clarify that my initial issue visible in the video above, i.e. grinding sound with lost frame sync, is not present at the moment. I almost did not have it anymore after opening the unit.
                        I'm currently fighting against the constantly periodic jitter in my converted videos (https://photos.app.goo.gl/49J1osrSN1EGXY9HA), which could be related anyway because both issues have a 4 frames period.

                        Yes David, it is plastic. Only the claw piece is metal (on a plastic mount).
                        I did not see anything strange when examining the pieces.
                        One question: is the large gear supposed to touch the cam? And is the small gear supposed to be pushed all the way down on the cam to keep it down, or is the spring enough for the cam to be in the correct position?
                        I'm asking because I think I could try detaching the gears a bit from the cam, to prevent any issues from their contact. I already did this with the large gear (raised it a bit) but nothing changed.

                        However I'm not sure how much time I will put into this, the converted movies are usable anyway (opposite to the results I got with the first grinding issue), and I'm now looking into ways to stabilize the frames based on the sprocket holes. This would probably give better results than what I could get even with a fixed mechanism. I found this suggestion by VideoFred in an old thread https://forum.doom9.org/showthread.p...08#post1584908 which I can use as a starting point. I could not find more recent info. Anyone knows of a better way? I'm just starting now with AviSynth scripts...

                        Stan I also wanted to finalise my understanding of the possibility to install Hawkeye in my unit: if I understand correctly, you think Hawkeye could fit. And I probably would have the same jitter issue I'm having now, right? (Well I have to deal with it anyway so it's not a negative point)
                        Any idea of the total cost it would have? I mean your board + camera + lens + mount or anything else if needed + shipping to Italy?
                        Thanks a lot.

                        Comment


                        • Marco, the Wolverine has 2 fundamental problems. Transport and image quality. The rest is very compact and clever.

                          I going to say I believe your problem is caused by gate tension. 2 things cause "jitter", as it is affectionately called, excessive gate tension and excessive takeup force.

                          Freidmann recently posted his solutions to these problems. He needed 250 grams of force to pull film thru the gate (with the claw down). That's crazy! I had tested mine between 30-40 grams and I lowered it to about half that. I no longer have any jitter issues.

                          Jitter can be stabilized in software. Fred's script has a stabilizer, the same (deshaker) is available for Virtual Dub and Vegas. Most editors have something. Vegas has multiple stabilizers. The cost is the image will be slightly cropped.

                          I don't remember noticing a rhythmic frame jump like yours. It is also difficult to explain the pattern. I think the claw struggles to advance the film and even makes a snapping sound when it descends that causes the film to recoil backwards slightly. That's how I imagine it.

                          It's also hard to explain grinding gears. These are large teeth and there is no play. There should be broken teeth. Yes, I know that no longer happens.

                          https://drive.google.com/file/d/1KRd...ew?usp=sharing

                          PS That is heavily compressed!

                          Here are some frames. The red lines hold constant to the moving frames.

                          Click image for larger version  Name:	m__000000.jpeg Views:	0 Size:	105.4 KB ID:	24927 Click image for larger version  Name:	m__000001.jpeg Views:	0 Size:	102.6 KB ID:	24926 Click image for larger version  Name:	m__000002.jpeg Views:	0 Size:	101.5 KB ID:	24922 Click image for larger version  Name:	m__000003.jpeg Views:	0 Size:	100.2 KB ID:	24924 Click image for larger version  Name:	m__000004.jpeg Views:	0 Size:	100.5 KB ID:	24923 Click image for larger version  Name:	m__000005.jpeg Views:	0 Size:	99.5 KB ID:	24925
                          Attached Files
                          Last edited by David Brown; January 04, 2021, 08:30 PM.

                          Comment


                          • 4 frame period is pointing to the large gear but the jitter indicates claw mechanism looseness or wear. The jitter is too high for Video Fred (option4) script to fix.
                            https://drive.google.com/file/d/1aVK...ew?usp=sharing
                            Tried the 01_E_Stabilisation_Only script as per your link.
                            Set the crop to the following values:
                            est_left=0 est_top=0 est_right=1300 est_bottom=850 #crop values for special Estimate clip

                            This grabs the area around the top sprocket hole. But still it does not work well. Maybe the sprocket hole should be more visible.
                            https://drive.google.com/file/d/1myA...ew?usp=sharing
                            The small gear on my unit is all the way down. If the cam is tight enough than I believe you can raise the small gear without affecting the cam operation.

                            The kit is $100. The BUC02 camera with the 12mm lens and holder is around $200.
                            16mm lens $105 and camera and holder is $165
                            UX178 camera and lens holder $348
                            UX226 camera with lens holder $379

                            Comment


                            • Hi you can maybe test this version of sprocket align avisynth script here https://8mmforum.film-tech.com/vbb/f...4416#post14416 it has work very good for me

                              here is my older version but it's slower https://8mmforum.film-tech.com/vbb/f...3022#post13022


                              in the avisynth script
                              dxmax=0,dymax=250


                              dxmax=horizontal
                              dymax=vertical

                              you can see I have disable dxmax=horizontal sprocket hole align Because I only need vertical sprocket hole align

                              you adjust there how many pixels it can different from the reference picture if it different more then the pixel number then it does not align that picture
                              Last edited by Mattias Norberg; January 05, 2021, 12:22 AM.

                              Comment


                              • Originally posted by Mattias Norberg View Post
                                Hi you can maybe test this version of sprocket align avisynth script here https://8mmforum.film-tech.com/vbb/f...4416#post14416 it has work very good for me
                                Thank you very much Mattias, I did a quick customization and test of your script and the result seems excellent!
                                Now I just need some more time to understand AviSynth to build my full processing chain .

                                Thank you Stan for the info about costs. The basic kit is not very expensive after all (especially if you can get Wolverine cheaply in first place). I suppose shipping costs need to be added yet. And maybe customs for delivery outside US.

                                Thanks for looking again into my issue. It seems strange to me that gate tension alone can cause such periodic issue. Maybe it is gate tension together with something else like gear deformation. But I have proof that it happens also without the belt on the take up reel, this short sample was acquired that way:
                                https://photos.app.goo.gl/8A4sB8rnzc5gGjZQ7

                                The previous sample which you used for your analysis was shared just to show the issue, but it is not very useful for attempting stabilization because it is a crop, too little of the sprocket holes is visibile. I'm now acquiring all movies in full zoom out so I have plenty of sprocket area, like in this sample above. And this is where I tested Mattias' script, with very good results.

                                PS: One more related discovery. I'm noticing now that the very first movies I converted after receiving my unit were much more stable. Not perfect, but rather good. So maybe the "jumping frames" started appearing after my initial issues with transport on some films, as if something got damaged by that.

                                Comment

                                Working...
                                X