is this the way to go decimation from 8mm fps to 25 fps?

[mammo1789]

Hey all

I'm still in "pain" with those old 8mm films and i want to try something is this correct way?

First the recordings are at 720x576 25 frames i lagarith. Some are black and white some are colors. I assume there are duplicates which according by the multidecimate there are frames that are 0, something apart and the threshold is 1 so it considers them duplicates after 2 pass i put to remove them. But the video goes and then it not goes to original video length but after let say 5 minutes the scene stays the same ( and original video continues and has more scenes rolling). What could be causing this. Then the frame rate drops to 9 something i used to do double deinterlace to get 18 fps and then go to fred script the results are ok but not satisfactory.

I think that after i delete the duplicates i can make picture sequence in vdub and get sequence of pictures ( is this better done when the picture is interlaced or should i deinterlace double frame first aka 18 something instead of 9 something ?)

The number of pictures will be much less then if i make picture sequence of full messed up 25i ( true 9 something video) and will be sorter duration probably.

So i want to put those pictures back in vdub and make 25 fps video. Now ( at least that what im thinking ) i have no duplicates and pure 25 fps video which can easily been applied with filters without problems and artifacts, instead of going to interpolation/blending of frames

Is this the way to go?


by the way the mfile.txt looks like this is there real or not duplicates


0 0.000000
46 0.964426
64 0.998613
142 0.726802
134 1.093158
123 0.736371
96 0.737658
67 0.926628
48 1.257202
30 0.863319
17 1.644897
8 1.342970
0 0.000000
1 1.776227
2 1.004069
3 1.032108
4 1.014321
5 0.813546
6 1.762951
7 0.832790
8 1.342970
9 0.841994
10 1.082640
11 2.061970
12 0.971160
13 1.166287
14 0.791811
15 0.853443
16 0.836291
17 1.644897
18 0.945742
19 1.502562
20 1.175774
21 0.700978
22 0.778105
23 1.964072
24 0.725666
25 1.826715
26 0.773778
27 0.870261
28 1.042539
29 2.057309
30 0.863319
31 1.595871
32 0.919496
33 1.001701
34 0.741020
35 2.137390
36 0.970060
37 1.360685
38 1.531868
39 0.730913
40 0.777487
41 1.907083
42 1.598107
43 1.192037
44 1.703135
45 1.040966
46 0.964426
47 0.763443
48 1.257202
49 1.053982
50 0.754709
51 1.471542
52 0.749854
53 0.720470
54 0.756636
55 1.367153
56 0.950641
57 0.736075
58 1.612662
59 0.992499
60 0.715164
61 0.972079
62 1.043799
63 1.810261
64 0.998613
65 1.473817
66 1.520404
67 0.926628
68 0.730288
69 0.729638
70 1.113920
71 1.649470
72 0.817247
73 0.929689
74 1.298074
75 0.829844
76 0.750007
77 1.092986
78 1.889830
79 1.863388
80 1.476549
81 3.681230
82 1.657126
83 4.319424
84 0.928779
85 0.714978
86 0.850467
87 1.314023
88 1.505976
89 0.654585
90 0.820697
91 1.345671
92 0.688776
93 0.658241
94 0.788019
95 0.662880
96 0.737658
97 1.306810
98 0.833294
99 0.753186
100 1.107107
101 1.220260
102 0.776870
103 0.710153
104 0.711215
105 0.754842
106 0.953736
107 0.918114
108 0.905114
109 0.624575
110 0.851980
111 1.015812
112 0.817347
113 0.858662
114 0.682462
115 0.669872
116 0.960633
117 1.158288
118 0.676712
119 0.807367
120 0.760844
121 0.818633
122 0.817552
123 0.736371
124 0.852092
125 0.711909
126 0.682289
127 0.732777
128 0.847143
129 0.708987
130 1.009646
131 0.621485
132 0.808935
133 1.151162
134 1.093158
135 0.998487
136 0.759189
137 0.711620
138 0.808306
139 0.677627
140 0.757966
141 0.832200
142 0.726802
143 0.802564
144 0.896034
145 0.707561
146 0.700007
147 1.073730
148 1.009334
149 0.897146
150 0.720295
151 0.852517
152 0.618332
153 1.019116
154 1.726897
155 0.993785
156 1.702304
157 8.636820
158 1.876127
159 2.663815

[manono]

Often dupe frames have metrics of one or above. What you've shown us looks to me to be from a relatively static scene. We'd need a sample to help figure this out and what the 'true' framerate is. But I doubt it's 9fps. Maybe 18fps or somewhere around there. If it's for PAL DVD you can encode at framerates as low as 16.67fps (as long as it's progressive) and apply DGPulldown after that. You don't have to hard telecine or insert dupe frames. And if they're from 8mm film, what makes you think it might be interlaced? By definition film is progressive. Unless, maybe, the fields don't line up properly. Do you see interlacing?

Anyway, sample please. One with steady movement.

[mammo1789]

And if they're from 8mm film, what makes you think it might be interlaced? By definition film is progressive. Unless, maybe, the fields don't line up properly. Do you see interlacing?

Thanks manono the interlacing comes from recording it is recorded by minidv camera ( but not recorded on camera but passtrough with cable to dvd recorder with rca cable from there to pc tv card "using only" the lens and recording with vdub 720x576 ( like would i have recorded any normal vhs tape) 25fps
test withot the decimation original

[manono]

Could you have picked a worse sample?

Anyway, sample please. One with steady movement.

[mammo1789]

Anyway, sample please. One with steady movement.

what do you mean by steady movement the file that i uploaded is the file that mfile is generated of that i posted. If you mean by no flicker they are all the same i will post another file

[manono]

I meant one with steady movement. With a static sample one can't determine the 'real' framerate or much of anything else. I expect it can be decimated fairly easily anyway (unless it's blended in which case a static sample is also useless), and you won't be needing MultiDecimate.

[mammo1789]

You mean like this ones

[2Bdecided]

I looked at test 2.avi.

You have poor focus, very strange levels, and horrendous flicker. Dupes are the least of your problems, but the flicker will prevent you from finding any. Where the same film frame appears on more than one video field/frame, it doesn't appear at the same brightness so is not an actual duplicate (in the sense that a dupe detector is looking for).

I think you still need film capturing help, rather than AVIsynth help, at this point. In my opinion (and I could be wrong), you need to find a way to capture each frame at least once with the same brightness. Try a different projector speed.

Hope this helps - but there must be threads here full of experts on this kind of thing.

Cheers,
David.

[manono]

Yes, much better, although the videos are so faint that it's hard to tell just exactly what's going on. Anyway, as near as I can tell the test 2.avi is 15fps (give or take) and can be handled something like this:

TFM()
TDecimate(Mode=0,Cycle=25,CycleR=10)

But, man, there's some bad flickering in there which, as 2Bdecided mentioned, can interfere with the duplicate detection and removal. test3 is different and it's blended in a changing pattern I don't know how to work with. Maybe someone else does. Otherwise I'd leave it at 25fps. Or capture it in a different way.

[mammo1789]

I looked at test 2.avi.

You have poor focus, very strange levels, and horrendous flicker. Dupes are the least of your problems, but the flicker will prevent you from finding any.

i know david but i use manual focus and yet it still bad

Where the same film frame appears on more than one video field/frame, it doesn't appear at the same brightness so is not an actual duplicate (in the sense that a dupe detector is looking for).

I don't know how to accomplish that.

I think you still need film capturing help, rather than AVIsynth help, at this point. In my opinion (and I could be wrong), you need to find a way to capture each frame at least once with the same brightness. Try a different projector speed.

I sure did also but the problem is that i have to work with the equipment that i have, also i don't have numbers on the projector so i assume the speeds of it should i try maybe couple of variants and see how it goes it will be very length process

Yes, much better, although the videos are so faint that it's hard to tell just exactly what's going on. Anyway, as near as I can tell the test 2.avi is 15fps (give or take) and can be handled something like this:

TFM()
TDecimate(Mode=0,Cycle=25,CycleR=10)

yes i know i sad in another tread that the film is very faint and i had to make histogram adjustments

Should i try deflicker before that

[jagabo]

Should i try deflicker before that

Yes.

[2Bdecided]

i don't have numbers on the projector so i assume the speeds of it should i try maybe couple of variants and see how it goes it will be very length process

if it's a variable speed control you can simply adjust it in real time, and look at the camcorder view finder (or on a TV monitor) to see when you've reached a speed that doesn't flicker.

I don't know if this will give you a video that it's possible to recover progressive frames from, but it should give you a video that's nicer to watch. If you're targeting a DVD, it's not too bad to leave it interlaced (not ideal with a progressive source, but as you say, you must work with what you have). If you want to upload it to the web / view it on a PC, then you need it progressive (easy enough to just deinterlace it), you really need it without blending (more difficult), and ideally you need it without duplicates (slightly more difficult still).

So maybe a simple interlaced transfer without flicker is the first thing to aim at?

A smarter solution might be a much slower transfer (turn the speed down further - don't burn the film though!), and a way of grabbing the good frames from what will then be a _very_ flickery result.

Someone else must have solved this problem, surely?

Cheers,
David.

[mammo1789]

Originally Posted by mammo1789
Should i try deflicker before that

Yes.

I tried it and it is much much better but still it is far from good, and cannot figure it out with more than one camera why the focus ( it is manual by the way) isn't sharp and on target i try all settings and it looks the sharpest that i can get, leaving that the projector might be "wrong" but outputet picture on the wall is very sharp crisp and detailed ( it looks even better than vhs because there is no chroma noise and such, so I am puzzled am I stupid or what

if it's a variable speed control you can simply adjust it in real time, and look at the camcorder view finder (or on a TV monitor) to see when you've reached a speed that doesn't flicker.

Hm that's also problem ether i have eyes problem or i don't see flicker on tv ( didn't tried viewfinder ) with the setup that gives the recorded video flicker, but when i play the recording or try it trough avisynth and such i get this sh..

I didn't find lot of post about 8mm transfers probably alot of guys forgot them especially setups ( most of the youtube is from Russians ) i didn't found simple scheme on how to build that kind of frame by frame transfer using web camera and modified mouse and software

[mammo1789]

clips