Why 7 or 15 repeated frames?

[sraposo]

2017-05-04: To save time and in order to get filtered information, I suggesto to jump to https://forum.videohelp.com/threads/383381-Why-7-or-15-repeated-frames#post2484966
.
.
.
Hi, All.

I´d like to get information, or at least clues, on the following case:
- ASF format video file (extension "wmv") generated by a vehicular video recorder;
- This video file has either 7 or 15 absolute frames with same image;
- softwares used on analysis: Vegas Pro, Windows Media ASF View 9 Series

The main question is: why, or what makes, a (*.wmv) video file gets (so many) replicated frames?

If you are interested in details, read the text below:
According Vegas Pro, we can find in this video either 7 or 15 absolute frames for each scenario. So, you have to move forward or backward sometimes 7, other times 15, absolute frames to get another scene. It doesn´t make sense to me that a video recorder spends processing time recording either 6 or 14 replicated frames of a same scene. In addition, why this variable number of repeated frames?

So, what is/are hypothesis regarding this frame repetition?

[poisondeathray]

Probably variable framerate recordings .

You can use mediainfo (view=>text) for more inforation

When you place a VFR video onto a constant framerate timeline like vegas, duplicates or blended duplicates will be inserted to make up the "FPS" . For example, if during a given second of real time, it records 6 images, there will be 5 duplicates per frame if you used a 30 FPS timeline. 10 duplicates if you used a 60FPS timeline

So they might not actually be "physically" recorded by your recorder, it might be the software inserting them.

VFR can be useful because during times of inactivity (it's the same image) , it doesn't make sense to record high framerates, you'd waste bandwidth and recording space. Or sometimes the recorder isn't capable of recording constant higher framerates (e.g. cheap recorders)

[sraposo]

Thanks, poisondeathray.

It´s not a VFR video, but a (average) 30 FPS.

This video is allegedly the original file created by a vehicular video recorder. It´s unlikely that such an equipment records using ASF format, although it´s not impossible, of course. I´m not also convinced that a vehicular video recorder would waste memory and processing time recording 7 or 15 identical frames.

So, I believe that this wmv file I´ve got is a file that was originally recorded in another format and converted to wmv, eventually setting some parameters that resulted in a frame-replicated file. Does it make sense? If not, what´s the reason? If YES, which parameters were defined to build such a file?

Check what Windows Media ASF View 9 Series reports:

ASF File (20160206135501.WMV)
File D:\Temp\20160206135501.WMV
EOF Position 161298668 ( 0x99D38EC )


File Properties Object (104 bytes)
Object ID 8CABDCA1-A947-11CF-8EE4-00C00C205365
Object Size 104 ( 0x68 )

Version 2
MMS ID AECA96BA-574D-4E3D-8912-2DC22A9B74F9
Total Size 161298668 ( 0x99D38EC )
Creation Time 2016-2-11 10:23:12.686
Packets 20160
Duration 36:56.193
Send Duration 36:54.175
Preroll 00:05.000
Flags 0x00000002
Broadcast 0
Seekable 1
Use Packet Template 0
Live 0
Reliable 0
Recordable 0
Unknown Data Size 0
Max Packet Size 8000 ( 0x1F40 )
Min Packet Size 8000 ( 0x1F40 )
Max Bitrate (bit/sec) 1680932


Data Object (not loaded) (161280050 bytes)
Object ID 75B22636-668E-11CF-A6D9-00AA0062CE6C
Object Size 161280050 ( 0x99CF032 )

MMS ID AECA96BA-574D-4E3D-8912-2DC22A9B74F9
Packets 20160
Alignment 1
Packet Aligment 1


Stream Properties Object [1] (114 bytes)
Object ID B7DC0791-A9B7-11CF-8EE6-00C00C205365
Object Size 114 ( 0x72 )

Stream Number 1
Version 1
Offset 0
Encrypted False
Security ID 1865290003 ( 0x6F2E1113 )
Stream Type Specific
Stream Type Audio Media
Format Tag 353
Channels 2
Samples / Seconds 44100
Average Bytes / Second 8005
Average Bitrate (bits/sec) 64040
Block Align 1487 ( 0x5CF )
Bits / Sample 16
Extra Data Size 10
Extra Data 0000: 00 88 00 00 0F 00 79 2E-00 00 y.
Error Concealment
Strategy Audio Spread
Span 1
Virtual Packet Length 1487 ( 0x5CF )
Virtual Chunk Length 1487 ( 0x5CF )
Silence Data 0000: 00


Extended Stream Properties Object [1] (88 bytes)
Object ID 14E6A5CB-C672-4332-8399-A96952065B5A
Object Size 88 ( 0x58 )

Stream Number 1
Start Time 00:00.000
End Time 00:00.000
Avg. Time / Frame 00:00.1853203
Avg. Frames / Second 5.40
Max. Object Size 1487 ( 0x5CF )
Avg. Data Bit Rate 64040
Max. Data Bit Rate 64040
Avg. Buffer Size 00:01.579
Max. Buffer Size 00:01.579
Avg. Initial Buffer Fullness 00:00.000
Max. Initial Buffer Fullness 00:00.000
Flags 0x00000002
Reliable 0
Seekable 1
No Clean Points 0
Resend Live Clean Points 0
Language Index 0


Stream Properties Object [2] (133 bytes)
Object ID B7DC0791-A9B7-11CF-8EE6-00C00C205365
Object Size 133 ( 0x85 )

Stream Number 2
Version 1
Offset 0
Encrypted False
Security ID 0
Stream Type Specific
Stream Type Video Media
Window Width 640
Window Height 480
Flags 2
Bitmap Info Header
biSize 44
Width 640
Height 480
Planes 1
Bits 24
Compression TEXT: WMV2
0000: 57 4D 56 32 WMV2
Image Size 0
X Pels / Meter 0
Y Pels / Meter 0
Colors Used 0
Colors Important 0
Extra Data Size 4
Extra Data 0000: F6 40 AD 00 @
Error Concealment
Strategy No Error Correction


Extended Stream Properties Object [2] (110 bytes)
Object ID 14E6A5CB-C672-4332-8399-A96952065B5A
Object Size 110 ( 0x6E )

Stream Number 2
Start Time 00:00.000
End Time 00:00.000
Avg. Time / Frame 00:00.0333333
Avg. Frames / Second 30.00
Max. Object Size 88526 ( 0x159CE )
Avg. Data Bit Rate 1600000
Max. Data Bit Rate 1600000
Avg. Buffer Size 00:05.000
Max. Buffer Size 00:05.000
Avg. Initial Buffer Fullness 00:00.000
Max. Initial Buffer Fullness 00:00.000
Flags 0x00000002
Reliable 0
Seekable 1
No Clean Points 0
Resend Live Clean Points 0
Language Index 0
Payload Extension Systems
System ID Data Size System Info Size System Info
Payload Extension System Sample Duration 2 0



Stream Bitrate Properties Object (38 bytes)
Object ID 7BF875CE-468D-11D1-8D82-006097C9A2B2
Object Size 38 ( 0x26 )
Stream Bitrates
Stream Number Bitrate
1 67271
2 1613661



Codec List Object (246 bytes)
Object ID 86D15240-311D-11D0-A3A4-00A0C90348F6
Object Size 246 ( 0xF6 )

Codec ID Reserved 2
Codecs
Name Description Type Format / FourCC Info Size Info
"Windows Media Audio 9.2" " 64 kbps, 44 kHz, stereo (A/V) 1-pass CBR" WMT_CODECINFO_AUDIO 353 2 0000: 61 01 a
"Windows Media Video V8" "" WMT_CODECINFO_VIDEO WMV2 4 TEXT: WMV2
0000: 57 4D 56 32 WMV2



Extended Content Description Object (168 bytes)
Object ID D2D0A440-E307-11D2-97F0-00A0C95EA850
Object Size 168 ( 0xA8 )

Attributes 3
Attributes
Index Name Stream Type Language Value
0 "WMFSDKVersion" 0 String 0 "12.0.7601.17514"
1 "WMFSDKNeeded" 0 String 0 "0.0.0.0000"
2 "IsVBR" 0 BOOL 0 False



Metadata Object (216 bytes)
Object ID C5F8CBEA-5BAF-4877-8467-AA8C44FA4CCA
Object Size 216 ( 0xD8 )

Attributes 4
Attributes
Index Name Stream Type Language Value
0 "IsVBR" 1 BOOL 0 False
1 "DeviceConformanceTemplate" 1 String 0 "L1"
2 "IsVBR" 2 BOOL 0 False
3 "DeviceConformanceTemplate" 2 String 0 "@"



Language List Object (39 bytes)
Object ID 7C4346A9-EFE0-4BFC-B229-393EDE415C85
Object Size 39 ( 0x27 )

Languages 1
Languages
Index Language
0 pt-br



Header Object (5254 bytes)
< show > Object ID 75B22630-668E-11CF-A6D9-00AA0062CE6C
Object Size 5254 ( 0x1486 )

Header Objects 7
Alignment 1
Architecture 2


Header Extension Object (4421 bytes)
< show > Object ID 5FBF03B5-A92E-11CF-8EE3-00C00C205365
Object Size 4421 ( 0x1145 )

Clock Type Reserved 1
Clock Size 6
Extended Header Size 4375 ( 0x1117 )


Compatibility Object (26 bytes)
< show > Object ID 26F18B5D-4584-47EC-9F5F-0E651F0452C9
Object Size 26 ( 0x1A )

Profile 2
Mode 1

[poisondeathray]

It´s not a VFR video, but a (average) 30 FPS.

"average" implies VFR

This video is allegedly the original file created by a vehicular video recorder. It´s unlikely that such an equipment records using ASF format, although it´s not impossible, of course. I´m not also convinced that a vehicular video recorder would waste memory and processing time recording 7 or 15 identical frames.

There are OEM equipment that record WMV/ASF.

You should find the make/model and check if you want to be sure

Equipment can record duplicate frames if the HW, buffer, or storage isn't fast enough . This is equivalent to a "dropped" frame, but just a "placeholder" of the previous frame in a CFR stream. Very common. Thus the content is VFR, but the recording is CFR. You can think of it as "VFR in CFR"

So, I believe that this wmv file I´ve got is a file that was originally recorded in another format and converted to wmv, eventually setting some parameters that resulted in a frame-replicated file. Does it make sense? If not, what´s the reason? If YES, which parameters were defined to build such a file?

Possibly, but usually original recordings will have metadata such as OEM, make, recording device etc...

It won't say something like "Windows Media Encoder 9.2 etc..." - that will suggest software encoding

If you wanted to make a file with duplicates , it's the same as described above. You start with a lower framerate and place it on a CFR timeline of higher number . In vegas you would disable resample to create pure duplicates (smart resample will "blend" frames). There are many other ways and software to do this as well (e.g. ffmpeg, avisynth etc...). But the one thing in common is a lower starting base frame rate . Also, it's possible to "wipe" metadata if to hide the software lineage

[sraposo]

""average" implies VFR"

> Ok!



"There are OEM equipment that record WMV/ASF. You should find the make/model and check if you want to be sure"

> It was already planned to do. While this information and others are not available, I´m trying to understand is or not original.



"Equipment can record duplicate frames if the HW, buffer, or storage isn't fast enough . This is equivalent to a "dropped" frame, but just a "placeholder" of the previous frame in a CFR stream. Very common. Thus the content is VFR, but the recording is CFR. You can think of it as "VFR in CFR""

> You mean that even 14 repeated frames is something expectable?



"Possibly, but usually original recordings will have metadata such as OEM, make, recording device etc..."

> YES! And such metadata doesn´t exist. But, I cannot state that this wmv file is not the original one because it lack such metadata. The equipment maker will be asked if the system records these complimentary information.



"If you wanted to make a file with duplicates , it's the same as described above. You start with a lower framerate and place it on a CFR timeline of higher number . In vegas you would disable resample to create pure duplicates (smart resample will "blend" frames). There are many other ways and software to do this as well (e.g. ffmpeg, avisynth etc...). But the one thing in common is a lower starting base frame rate . Also, it's possible to "wipe" metadata if to hide the software lineage"

> My last doubt is: does Vegas Pro exhibits the video with replicated frames (7 or 15 frames with same image), but these repeated frames don´t exists in the file? Or does the wmv file has truly all those replicated frames?

[sraposo]

poisondeathray,

According timestamp, the wmv file presents 2 images per second. So, it would be reasonable that the file had a FPS = 2, with no replicated frames. Is that right?

Since this wmv file is 30FPS (average), 15 frames are needed (most of the images occupies 15 frames. Sometimes it´s only 7) for each scene. So, I suppose, these 15-frame run for each scene are really recorded in the wmv file, and it´not some "trick" that Vegas Pro does to comply to the 30FPS set in the file header.

In addition: this wmv file has a 17-second extent missing. After scene with "13:55:44" timestamp, it comes a new scene with a "13:56:01" timestamp, as a probable result of a cut.

So, it seems that this wmv file is a result of edition of another video (probably the original one) and it (the wmv) was recorded with a FPS = 30.

Does it make sense?

[poisondeathray]

"Equipment can record duplicate frames if the HW, buffer, or storage isn't fast enough . This is equivalent to a "dropped" frame, but just a "placeholder" of the previous frame in a CFR stream. Very common. Thus the content is VFR, but the recording is CFR. You can think of it as "VFR in CFR""

> You mean that even 14 repeated frames is something expectable?

I wouldn't say "expected", but adverse conditions can predispose recordings like that as well (e.g. maybe very hot in a car , or too cold)

But I wouldn't expect ONLY 7 and 15 repeats if it was a HW issue; I would expect some 6, some 12 etc... ie. some variation

"Possibly, but usually original recordings will have metadata such as OEM, make, recording device etc..."

> YES! And such metadata doesn´t exist. But, I cannot state that this wmv file is not the original one because it lack such metadata. The equipment maker will be asked if the system records these complimentary information.

Check again with mediainfo , and also a hex editor . I would say the majority will say something , at least a hardware encoder ID

"If you wanted to make a file with duplicates , it's the same as described above. You start with a lower framerate and place it on a CFR timeline of higher number . In vegas you would disable resample to create pure duplicates (smart resample will "blend" frames). There are many other ways and software to do this as well (e.g. ffmpeg, avisynth etc...). But the one thing in common is a lower starting base frame rate . Also, it's possible to "wipe" metadata if to hide the software lineage"

> My last doubt is: does Vegas Pro exhibits the video with replicated frames (7 or 15 frames with same image), but these repeated frames don´t exists in the file? Or does the wmv file has truly all those replicated frames?

Yes, when you do it in the matter described, they will be encoded into the file . ie. "hardcoded" duplicates .

According timestamp, the wmv file presents 2 images per second. So, it would be reasonable that the file had a FPS = 2, with no replicated frames. Is that right?

Yes, that should be correct.

Instead of vegas, you can check with something like avisynth (e.g ffms2) to index the file, so you only see the actual encoded images . But that doesn't necessarily indicate the heritage of the video. An authentic original recording may very well be timestamp VFR (timestamp VFR has no actual encoded duplicates; as opposed to the "VFR in CFR" which hardcoded duplicates, the "placeholders" mentioned earlier)

Also, the actual framerate might be 2 or it might change. If it changes that's VFR by definition. If it were 2 FPS constant the whole way through , no deviation, that's CFR

Since this wmv file is 30FPS (average), 15 frames are needed (most of the images occupies 15 frames. Sometimes it´s only 7) for each scene. So, I suppose, these 15-frame run for each scene are really recorded in the wmv file, and it´not some "trick" that Vegas Pro does to comply to the 30FPS set in the file header.

You can check by indexing the actual ("physically encoded") images with the avisynth ffms2 method

In addition: this wmv file has a 17-second extent missing. After scene with "13:55:44" timestamp, it comes a new scene with a "13:56:01" timestamp, as a probable result of a cut.

So, it seems that this wmv file is a result of edition of another video (probably the original one) and it (the wmv) was recorded with a FPS = 30.

Does it make sense?

Difficult to say, because if you edit out a segment, you can do with tools that pass through the original stream unaltered (ie. "direct stream copy" ; no quality loss when using GOP boundary cuts) . So with that information alone, it might still be original, just a segment edited out. An example of a tool that can do this for wmv/asf streams is asfbin

However, if you decompress and recompress a video, then it's definitely no longer 1st generation . And if it's using WMV, then definitely quality lost.

[sraposo]

Hi!



"Check again with mediainfo , and also a hex editor . I would say the majority will say something , at least a hardware encoder ID"

> I´ve already done that. Up to this moment, Windows Media ASF View 9 Series has been the most detailed software. That bunch of data I posted before is only a fraction of what Windows Media ASF View 9 Series can provide



"Instead of vegas, you can check with something like avisynth (e.g ffms2) to index the file, so you only see the actual encoded images "

> I´ve tried other editors and Vegas Pro was the only that showed all the replicated frames! Other editors only showed much less replicated frames, even 1 frame per scene!
But I´ll follow your suggestion and try avisynth!



If there is a 2-FPS system, it would be enough to make such a file that presents a new scenario each 0,5s (2 Hz). But, as dash cam recorders on the market works with at least 15FPS (most at 30FPS), so image refresh would happen each 67ms, much faster than that 500ms that can be seen on that WMV file I mentioned. With a 30FPS specification in the file header and 15 frames per scene, you see a movie that changes scenes each 0.5, or 2 scenes per second. That´s makes sense.

But what DOESN´T make sense is a movie (allegedly original) that shows 2 images per second has been recorded with 14 repeated frames (1 frame with a useful image and 14 more frames with the same image), this way wasting media room and processing time.

Applying what you commented previously, I made some experiments that resulted in a movie that has a image presentation rate of 2 FPS and multiple frames with same image:
1) took a video and created another with a 2 FPS rate;
2) cut off a piece of this 2 FPS video;
3) saved it with a set of parameters that resulted in a 25 or 30 FPS
The resulting file has FPS/2 replicated frames and it changes scenes at a 0.5 interval.

Thanks to your priceless help, I could understand a little bit more about this and I maybe I´ve had some progress: I´ve got 2 hypothesis, and it considers that this WMV file is not the original one:
1) the original video was 15, 30, whatever (much) more than 2 FPS, had a extent cut off and finally was saved to a file taking frames with a 14-frame interval. In other words, taking 1 frame and skipping 14. Most dash cam records on the market records at a 30FPS rate, some at 15FPS, others at more than 30FPS. This is technically possible, but I think it´s not likely;
2) the original recording was really done with a 2 FPS rate (although I don´t know if some dash cam recording system allows such recording rate) and a new file was created after the original has been edited and saved with a different quality, that increased FPS (thus, resulting the replicated frames).

[poisondeathray]

> I´ve tried other editors and Vegas Pro was the only that showed all the replicated frames! Other editors only showed much less replicated frames, even 1 frame per scene!
But I´ll follow your suggestion and try avisynth!

That suggests actual 2FPS recording, if "other" editors showed 1 frame per scene . Vegas is probably adding the extra frames as described to make up the CFR timeline. If you put it on a 2FPS timeline (in the vegas project properties), that 2FPS content would only show unique frames, right ? If you used a 10FPS project you would have strings of 5 etc...

If there is a 2-FPS system, it would be enough to make such a file that presents a new scenario each 0,5s (2 Hz). But, as dash cam recorders on the market works with at least 15FPS (most at 30FPS), so image refresh would happen each 67ms, much faster than that 500ms that can be seen on that WMV file I mentioned. With a 30FPS specification in the file header and 15 frames per scene, you see a movie that changes scenes each 0.5, or 2 scenes per second. That´s makes sense.

Yes, 30FPS is the most common

But what DOESN´T make sense is a movie (allegedly original) that shows 2 images per second has been recorded with 14 repeated frames (1 frame with a useful image and 14 more frames with the same image), this way wasting media room and processing time.

Yes, but you haven't confirmed this either way yet. Your observations with "other editors" suggests they are not actually encoded.

1) the original video was 15, 30, whatever (much) more than 2 FPS, had a extent cut off and finally was saved to a file taking frames with a 14-frame interval. In other words, taking 1 frame and skipping 14. Most dash cam records on the market records at a 30FPS rate, some at 15FPS, others at more than 30FPS. This is technically possible, but I think it´s not likely;
2) the original recording was really done with a 2 FPS rate (although I don´t know if some dash cam recording system allows such recording rate) and a new file was created after the original has been edited and saved with a different quality, that increased FPS (thus, resulting the replicated frames).

If you have a dashcam recording native WMV, it's going to be a very old model (very early generation) . You won't find any in North America , except in museums or garage sales. But cameras using native WMV do exist. Modern models all use AVC compression, and you're right , typically 30FPS for dashcams

But why does it have to be an offical actual "dashcam" ? Is there some other information like a "garmin" overlay or something ? Why can it not just some portable device, a phone, a camera, even a webcam - just mounted on the dash ? Because modern devices like phones almost all record timecode VFR (but again mostly AVC) . But you will almost never get exactly strings of 7 and 15 repeats. If there is no "random" variation, that's highly suspicious of not being 1st generation for those .

Also check with hex editor. There can be hidden tags and information not revealed by conventional software

[sraposo]

"That suggests actual 2FPS recording, if "other" editors showed 1 frame per scene . Vegas is probably adding the extra frames as described to make up the CFR timeline. If you put it on a 2FPS timeline (in the vegas project properties), that 2FPS content would only show unique frames, right ? If you used a 10FPS project you would have strings of 5 etc..."

"Yes, but you haven't confirmed this either way yet. Your observations with "other editors" suggests they are not actually encoded."

> The property windows doesn´t allow to change FPS (see below). If I did the right thing, I set Vegas Pro to use a costumized Timecode and "absolute frames" and that repeated frames pattern was kept. So, I think I can realize that all those repeated frames are real.




I´ve just noticed that I presented some information in a not accurate way: the video was made by a 4-channel dash cam recorder and each cam image is simultaneously shown in the usual way.
Cam 1 is the upper left image, Cam 2 is the upper right one, Cam 3 is the lower left one and Cam 4, the lower right part, is missing.
Cam 1 and Cam 3 are refreshed at same time and stay the same for 15 frames. The same with Cam 2 and Cam 4, but there´s usually a 7-frame interval between Cam1/Cam3 and Cam2/Cam4 refreshments. I said "usually" because sometimes this 15-frame repetition doesn´t happen, but a 7-frame one. So, sometimes all 4 cam images are simultaneously refreshed.



"But why does it have to be an offical actual "dashcam" ? Is there some other information like a "garmin" overlay or something ? Why can it not just some portable device, a phone, a camera, even a webcam - just mounted on the dash ? Because modern devices like phones almost all record timecode VFR (but again mostly AVC) . But you will almost never get exactly strings of 7 and 15 repeats. If there is no "random" variation, that's highly suspicious of not being 1st generation for those ."

> The previous paragraph answers that. It´s a four-channel vehicular video recorder.




"Also check with hex editor. There can be hidden tags and information not revealed by conventional software"

> I´ve already done that. No metadata found, unfortunately.

[poisondeathray]

> The property windows doesn´t allow to change FPS (see below). If I did the right thing, I set Vegas Pro to use a costumized Timecode and "absolute frames" and that repeated frames pattern was kept. So, I think I can realize that all those repeated frames are real.

Not the event properties

If you change vegas project properties, the timeline settings will change. If you set it to 2 fps, 2fps sections will be unique with no duplicates . If you set it to 60 FPS, now you have 30 repeat frame sections. Are those frames real ? Absolutely not. So vegas is useless for making that determination , I hope that make sense

You want to see individual encoded frames, not software interpolated frames; vegas is wrong tool because your project properties settings will alter what you have



I'm not familiar with those 4 way models, is that one 640x480 image split into 4 ? not very useful resolution .

You need to find the make/model of the recorder and check with the manufacturer

[sraposo]

"So vegas is useless for making that determination , I hope that make sense"
"You want to see individual encoded frames, not software interpolated frames; vegas is wrong tool because your project properties settings will alter what you have"

> Now I see what you meant. That was an unexpected effect to me, regarding video editors. Yes, Vegas Pro is not the correct tool.
I tried ASFBin 1.8 and I could see the encoded frames. It´s a 4 FPS video, indeed.
The nearest information regarding this provided by Windows Media ASF View 9 Series is "Avg. Frames / Second 5.40 ", that also presents "Avg. Frames / Second 30.00". I´ve taken a second look in MediaInfo and something similar is informed: "Frame rate: 4.000 fps" and "Nominal frame rate: 30.000 fps"
What means this "Nominal frame rate: 30.000 fps" if the file is really 4 FPS?



"I'm not familiar with those 4 way models, is that one 640x480 image split into 4 ? not very useful resolution"

> Yes, a 640x480 area splited into 4. Not a great definition, but it may be enough in certain cases.



"You need to find the make/model of the recorder and check with the manufacturer"

> Unfortunately it will not be possible for while, but these informations will be asked. If they come, things will become a little easier.

At this moment, that wmv video is the only material and I´m trying to check its authenticity.

[poisondeathray]

> Now I see what you meant. That was an unexpected effect to me, regarding video editors. Yes, Vegas Pro is not the correct tool.
I tried ASFBin 1.8 and I could see the encoded frames. It´s a 4 FPS video, indeed.
The nearest information regarding this provided by Windows Media ASF View 9 Series is "Avg. Frames / Second 5.40 ", that also presents "Avg. Frames / Second 30.00". I´ve taken a second look in MediaInfo and something similar is informed: "Frame rate: 4.000 fps" and "Nominal frame rate: 30.000 fps"
What means this "Nominal frame rate: 30.000 fps" if the file is really 4 FPS?

Nominal as reported by mediainfo is what the header reports the frame rate as .

Anytime you have an "average" frame rate reported that imples VFR . Timecodes keep everything in sync (even though you probably don't have audio here, I mean in the general case), and the proper duration

So when you play it back in a media player, the frame rate is "30" . 30 images are displayed per second. But individual frames are displayed for different durations based on timecodes. Some frames are repeated to "make up" that 30 images. That is VFR.

(And when you "watch" on a 60Hz display, those 30 fps are yet again displayed for twice the duration each, because that "30" is made into "60")

[sraposo]

"Nominal as reported by mediainfo is what the header reports the frame rate as . Anytime you have an "average" frame rate reported that imples VFR . Timecodes keep everything in sync (even though you probably don't have audio here, I mean in the general case), and the proper duration"

> OK, it´s very clear "average" is a implicit synonym of VFR. Curiously (to me), all the extent (the initial portion) that shows the bus (and the targets pointed by the other 2 cams) in movement, you find 4 FPS. After the bus crash, there is a 35 minute portion with the bus stopped. So, it seems strange (to me) the VFR characteristic in a movie where there would have stretches with more than 4FPS that would have occurred right on portions with low or no activity.
Anyway, this nominal FPS data won´t be useful as an evidence or a clue to lead to something relevant.

[jagabo]

It's normal for Microsoft's WMV encoders to reduce the frame rate when there is too much action for the requested bitrate. The theory being it's better to deliver fewer "clean" frames per second than many "dirty" frames per second.

[sraposo]

Hi, Jagabo. Thanks for engaging this discussion.

You meant that 30FPS (nominal frame rate) would be the "intented" FPS but 4fps would have been the rate that the system managed to record providing better material?

[sraposo]

Hi all.

At this moment, I think that a video editor that showed only encoded frames (nothing like Vegas Pro ) and exhibited low level data related to each frame( index data, payload data, ASF format structure fields and others) might unveil something useful. To use a hex editor would demand a huge effort to interpret all those fields, so it could be useful to check only little portions of bytes in a very low level, if really needed.

ASFBin and Windows Media ASF View do only part of these functions each one. So, does anyone know a software that presents internal data related to each frame?

[jagabo]

ffprobe, which comes with ffmpeg, can give you a list of frames and their properties.

ffprobe.exe -print_format json -show_frames input.asf > output.txt

[sraposo]

Thanks, Jagabo, for your suggestion.

You´re right, ffprobe shows a lot of interesting information, but I am still looking fot a video editor or player that shows inner data related to the current frame.

[jagabo]

If you set up ffdshow as the decoder for the video you can enable its on screen display:


[Attachment 41480 - Click to enlarge]

The information will be written onto the video.

[sraposo]

YES, Jagabo!

That´s EXACTLY what I was looking for!

Thanks a lot!

[sraposo]

Sirs,

After deeper checking and compiling information that was posted during this thread, that´s what is relevant:
1- ASF format video file (wmv file) made by a 4-channel dash cam recorder. Unknown maker, model and settings
2- frame rate: 4 FPS. That´s what happens when video is played
3- nominal frame rate: 30FPS
4- from frame 0 till 87 and from 172 on, 2 cams are updated at each frame, alternatively to the other 2 cams
5- from frame 88 till 171, a pair of even-numbered/odd-numbered (e.g 88/89, 90/91,...) shows the same image, changing all four cam images at a time
6- frame 149 is missing
7- there´s a 16-second jump in time showed by timestamps on frames 171 and 172

The mission is check authenticity of this file.

The main fact that catch anyone´s attention is item 7, that strongly suggests a cut off, but that is allegedly due to a recording system failure, according the person who provided the video.

It´s quite curious what happens in itens 5 and 6

In a small degree, item 3 doesn´t seem to comply to the average frames per second that are seen during playing.

[sraposo]

Complimentary info:

8- Field "Creation Time" in "File Properties Object": 2016-2-11 10:23:12.686 (five days after accident). This field, according Advanced Systems Format (ASF) Specification, Revision 01.20.03, edited by Microsoft Corporation (December 2004), "specifies the date and time of the initial creation of the file. The value is given as the number of 100-nanosecond intervals since January 1, 1601, according to Coordinated Universal Time (Greenwich Mean Time).(...)"

It´s quite curious what happens in itens 5, 6 and 8.

[poisondeathray]

also look at exiftool

compare file modification date/time, file access date/time, file creation date/time with the creation date

file creation date/time will change if you copy or move the file to a new system/computer . This is normal. But "creation date" should be the same as original, and not change - that should be the original system date/time of the original recording

If the system time of the recording is shifted (e.g. maybe it wasn't setup correctly, or the wrong date/time) , then all recordings should be shifted by the same amount . See if you can get other recordings from known dates from the same device. This also allows you to compare the WMV file/structure (if it's even WMV to begin with)