VideoHelp Forum

+ Reply to Thread
Page 3 of 4
FirstFirst 1 2 3 4 LastLast
Results 61 to 90 of 115
Thread
  1. The "newspaper/menu" has personal information ?

    Crop to the region of interest (the "newspaper/menu") and export more than 5 frames before and after (at least 11 or more frames) using a lossless codec

    Or mask out the area, then export
    Quote Quote  
  2. Originally Posted by poisondeathray View Post
    The "newspaper/menu" has personal information ?

    Crop to the region of interest (the "newspaper/menu") and export more than 5 frames before and after (at least 11 or more frames) using a lossless codec

    Or mask out the area, then export
    No because infidel naysayers will say I cheated, and it is suspicious blah blah... I willl shot a different new video.
    Quote Quote  
  3. Originally Posted by Truthler View Post
    Originally Posted by poisondeathray View Post
    The "newspaper/menu" has personal information ?

    Crop to the region of interest (the "newspaper/menu") and export more than 5 frames before and after (at least 11 or more frames) using a lossless codec

    Or mask out the area, then export
    No because infidel naysayers will say I cheated, and it is suspicious blah blah... I willl shot a different new video.
    But you're giving people ammunition to make those claims -


    If you post an example, but nobody can reproduce it - not even with Neat Video - how can anyone verify the claims ?

    This is basic scientific method 101 .

    Why would a person believe those claims ? There is no valid "proof" unless results are reproducible (or at least similar results, because you might sample a slightly different region in Neat Video)

    eg . How does someone distinguish whether or not a marketing company didn't just make up some fake results ?


    And now you're switching the video, making it look suspicious - you're giving people reason to doubt
    Quote Quote  
  4. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by poisondeathray View Post
    This is basic scientific method 101 .
    And now you're switching the video, making it look suspicious - you're giving people reason to doubt
    He's just trolling. But we all already knew that. Some of you keep humoring him, but he'll continue to move goal posts, make BS claims, etc.

    - He plays with HD video, using OBS/whatever. Great.
    - Then he uses easyware to process. Great.
    - But then he tries to convince us that his basic understanding of video is more genius (Wile E. Coyote) than any of us could possibly ever understand. Not great. I never understand the motivations of trolls.

    Originally Posted by Truthler View Post
    Don't forget: I started digitalizing VHS 2 years before you.


    By your own admission, you were teenager capturing 320x240 with an infamous BT8x8 chipset (ie, crap quality captures). Meanwhile, I was dabbling with SGI MediaBase (T1 bandwidth MPEG) the year prior, and I wasn't a kid. And I started with analog video long before that, seemingly back when you were still learning your ABCs and 123s.
    Last edited by lordsmurf; 13th Sep 2021 at 13:57. Reason: clarifications
    Quote Quote  
  5. Originally Posted by poisondeathray View Post
    Originally Posted by Truthler View Post
    Originally Posted by poisondeathray View Post
    The "newspaper/menu" has personal information ?

    Crop to the region of interest (the "newspaper/menu") and export more than 5 frames before and after (at least 11 or more frames) using a lossless codec

    Or mask out the area, then export
    No because infidel naysayers will say I cheated, and it is suspicious blah blah... I willl shot a different new video.
    But you're giving them ammunition to make those claims -


    If you post an example, but nobody can reproduce it - not even with Neat Video - how can anyone verify the claims ?

    This is basic scientific method 101 .

    Why would a person believe those claims ? There is no valid "proof" unless results are reproducible (or at least similar results, because you might sample a slightly different region in Neat Video)

    eg . How does someone distinguish whether or not a marketing company didn't just make up some fake results ?
    BEcause I will use a long video, During the video shot I will slowly raise the gain/iso to the max, raising slowly shutter speed etc.... So it will be visible that all of these things are happening during the shot.
    Quote Quote  
  6. Originally Posted by lordsmurf View Post
    Originally Posted by poisondeathray View Post
    This is basic scientific method 101 .
    And now you're switching the video, making it look suspicious - you're giving people reason to doubt
    He's just a f'ing troll. But we all already knew that. Some of you keep humoring him, but he'll continue to move goal posts, make BS claims, etc.

    Originally Posted by Truthler View Post
    Don't forget: I started digitalizing VHS 2 years before you.


    By your own admission, you were teenager capturing 320x240 with an infamous BT8x8 chipset (ie, crap quality capture/encode). Meanwhile, I was dabbling with SGI MediaBase (T1 bandwidth MPEG) the year prior, and I wasn't a kid. First on an SGI workstation, then the new Mac OS8. And I started with analog video long before that, back when you were still learning your ABCs and 123s.
    If you don't have proofs and arguments you start ad-hominem personal attacks. Personal attack is considered as a logical fallacy.

    You were a TV repairer as I mentioned. It was better to record in 320*240px in 1998 than just dreaming about it.... :P You made such fantastic claims and statements that there are no video engineers.... Just ask Selur about the exact name of his university degree. And Selur is not a teenager... so he is the living example among Videohep users that such university degree existed even long-long time ago. It is not my fault that you did not hear about that, and you remained a TV repairer.
    Last edited by Truthler; 13th Sep 2021 at 13:49.
    Quote Quote  
  7. Originally Posted by Truthler View Post

    BEcause I will use a long video, During the video shot I will slowly raise the gain/iso to the max, raising slowly shutter speed etc.... So it will be visible that all of these things are happening during the shot.
    That's a good idea too -


    But your "newspaper" example gets thrown out in many people's minds, because you provided nothing verifiable

    In North America, there is a saying "shooting yourself in the foot". You post an nice example to demonstrate some observations (it is a really good result, much improved) - but it's counter productive and working against you

    But there are a few people in this thread that use current version of Neat Video - they can verify this example. That adds weight to those claims

    If that's not "enough" - The Neat Video demo has some limitations (such as 1280x720 area, short duration, and watermark) ; but it's good enough to evaluate a region of interest on the newspaper - that's good enough "proof" for anybody, because anybody can reproduce those results, if they are valid.

    A small cropped region of interest is also good enough to see if some "open source" alternative can produce similar results . You "think" that open source cannot do it, why not give open source a chance, or maybe open source will be "defeated" . This is your big chance Selur is willing to take up the challenge .
    Quote Quote  
  8. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by Truthler View Post
    Personal attack is considered as a logical fallacy.
    (1) I don't think you understand what a logical fallacy is.
    (2) You're the one saying nonsense, as it concerns my video past.

    You were a TV repairer as I mentioned.
    I've never repaired TVs.

    You made such fantastic claims and statements that there are no video engineers.... .
    I don't think you understand the various roles and niches of video, and I never said "there are no video engineers" (as there are certainly broadcast engineers).
    Quote Quote  
  9. You are one high quality troll, Truthler
    Quote Quote  
  10. Originally Posted by lordsmurf View Post
    Originally Posted by Truthler View Post
    Personal attack is considered as a logical fallacy.
    (1) I don't think you understand what a logical fallacy is.
    (2) You're the one saying nonsense, as it concerns my video past.

    You were a TV repairer as I mentioned.
    I've never repaired TVs.

    You made such fantastic claims and statements that there are no video engineers.... .
    I don't think you understand the various roles and niches of video, and I never said "there are no video engineers" (as there are certainly broadcast engineers).
    You believed that there are no broadcast / video related Engineers or anything similar.... You made a nonsense statement, and as I named such a person, you flee to denial.

    Personal attack, (ad Hominem) is a logical fallacy https://thebestschools.org/magazine/15-logical-fallacies-know/
    Using ad hominem, the person usually tries to hide the fact that he/she has no agruments and proofs in the debate.
    Quote Quote  
  11. The noisy contest video will be shot until sunday.
    Quote Quote  
  12. Good to hear, take your time to create/select the video that you will think offers a valid/fair basis for a comparision.

    Cu Selur
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  13. Originally Posted by Selur View Post
    Good to hear, take your time to create/select the video that you will think offers a valid/fair basis for a comparision.

    Cu Selur
    Let's chose your scenery for the noisy video, where I had to shot the video.

    (Turn around: the left mouse button and drag :

    Press esc for aerial view.


    Scenery 1.

    https://www.google.com/maps/@47.5017329,19.0344516,3a,60y,301.63h,124.83t/data=!3m6!1e...7i13312!8i6656

    Scenery 2.

    https://www.google.com/maps/@47.4961275,19.0403659,3a,75y,48.92h,101.74t/data=!3m6!1e1...7i13312!8i6656


    Scenery 3.

    https://www.google.com/maps/@47.5065278,19.0479252,3a,75y,298.26h,97.84t/data=!3m6!1e1...7i13312!8i6656

    Scenery 4.

    https://www.google.com/maps/@47.5147631,19.0777698,3a,75y,30.59h,92.93t/data=!3m6!1e1!...7i13312!8i6656

    Scenery 5.

    https://www.google.com/maps/@47.5052704,19.0634909,3a,75y,147.09h,102.84t/data=!3m6!1e...7i16384!8i8192

    Scenery 6.

    https://www.google.com/maps/@47.5150794,19.0435785,3a,60y,189.41h,85.79t/data=!3m6!1e1...7i13312!8i6656

    Scenery 7.

    https://www.google.com/maps/@47.4932057,19.05582,3a,75y,218.32h,109.22t/data=!3m6!1e1!...7i16384!8i8192

    Scenery 8.

    https://www.google.com/maps/@47.4997723,19.0552559,3a,75y,340.87h,92.68t/data=!3m6!1e1...7i16384!8i8192

    Scenery 9.

    https://www.google.com/maps/@47.4984837,19.0700857,3a,60y,82.4h,119.03t/data=!3m6!1e1!...7i16384!8i8192

    Scenery 10.

    https://www.google.com/maps/@47.4867931,19.0481264,3a,75y,24.11h,76.29t/data=!3m8!1e1!...!7i8192!8i4096

    Scenery 11:

    https://www.google.com/maps/@47.5096774,19.0570425,3a,75y,20.56h,95.22t/data=!3m6!1e1!...7i13312!8i6656

    Scenery 12:

    https://www.google.com/maps/@47.5097223,19.0700201,3a,60y,169.13h,91.56t/data=!3m6!1e1...7i16384!8i8192

    Scenery 13:

    https://www.google.com/maps/@47.5044096,19.0503801,3a,75y,244.15h,89.35t/data=!3m8!1e1...!7i9216!8i3588

    Scenery 14.

    https://www.google.com/maps/@47.498329,19.0407381,3a,75y,70.76h,96.45t/data=!3m6!1e1!3...7i16384!8i8192
    Quote Quote  
  14. I got no preference, but may be someone else has.
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  15. Originally Posted by Selur View Post
    I got no preference, but may be someone else has.
    Let's choose one, it is not a great effort to watch them.
    Quote Quote  
  16. Like I wrote: I have no preference, they all the scenaries have potentially lots of details and light differences. Since it's not about super resolution or similar but denoising, I vote for Scenery 3 (that's what came up when rolling two 6 sides dices).
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  17. 3 because why not
    Quote Quote  
  18. Originally Posted by Selur View Post
    Like I wrote: I have no preference, they all the scenaries have potentially lots of details and light differences. Since it's not about super resolution or similar but denoising, I vote for Scenery 3 (that's what came up when rolling two 6 sides dices).
    Hello Selur!

    Here is the pre contest file (5 Gbyte), it contains pre-test. https://sendgb.com/loxryMsbmjY

    It was recorded in lossless codec with CAM LINK 4K.

    You can see on the bottom of the picture the original camera tech data of the Sony FDR-AX700 camcorder, such as the IRIS GAIN Shutter speed (in this order), and you can see as I manually changed them during the recording to get as much as noise as possible.


    Of course, I will also shot the promised Scene 3 (Parliament Building) later.

    You can send/upload your own version of the denoised file via the sendgb.com (since Videohelp storage is very very limited) , because the output file must be in lossless format.
    Last edited by Truthler; 17th Sep 2021 at 09:43.
    Quote Quote  
  19. pre contest file?

    It was recorded in lossless codec vide CAM LINK 4K.
    Okay it's UT Video 4:2:0 content (ULH0).


    -> Video seems kind of useless with all the camera overlay info,... (through the whole clip)
    (as a side note: what are all the red dots for? or are those some kind of noise/defect you encounter?)

    Cu Selur
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  20. Originally Posted by Selur View Post
    pre contest file?

    It was recorded in lossless codec vide CAM LINK 4K.
    Okay it's UT Video 4:2:0 content (ULH0).


    -> Video seems kind of useless with all the camera overlay info,... (through the whole clip)
    (as a side note: what are all the red dots for? or are those some kind of noise/defect you encounter?)

    Cu Selur
    I turned on original camera overlay, to procve that the noise is generated by the camera with 10000 ISO and high shutter speed (1/4000)

    PEAKING ASSIST for focus.
    Read about it here:
    https://helpguide.sony.net/cam/1710/v1/en/contents/TP0001492439.html

    But why didn't you chose frames where peaking assist are not or less present?
    Quote Quote  
  21. But why didn't you chose frames where peaking assist are not or less present?
    Jumped through the source to see whether the on-screen-display content was through out the whole clip (and it was) and then jumped back to a frame where the 'peaking assist' dots were good visible to be sure it was clear what I was refering to when I asked about the 'red dots'.
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  22. Originally Posted by Selur View Post
    But why didn't you chose frames where peaking assist are not or less present?
    Jumped through the source to see whether the on-screen-display content was through out the whole clip (and it was) and then jumped back to a frame where the 'peaking assist' dots were good visible to be sure it was clear what I was refering to when I asked about the 'red dots'.
    A more difficult video. Only 234 Mbyte. Please uncover the handwritten (ballpointpen) info in the result.

    https://sendgb.com/Os2cvirX7QL
    Last edited by Truthler; 17th Sep 2021 at 10:45.
    Quote Quote  
  23. Please uncover the handwritten (ballpointpen) info in the result.
    Not totally sure I understand what you mean by 'uncover', do you mean something like this?


    Cu Selur
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  24. Originally Posted by Selur View Post
    Please uncover the handwritten (ballpointpen) info in the result.
    Not totally sure I understand what you mean by 'uncover', do you mean something like this?


    Cu Selur
    Don't hide anything. Do you know the exact frame number what you selected to post?
    Quote Quote  
  25. No, no clue what frame I took the screen shot.
    So you want to remove that writing? Okay that sounds like one should creating a mask, remove the content and then apply some inpainting technique.
    (not really denoising)
    I have never tried that, so this will take some time. but it should be similar to logo removal, so jagabo und some of the other users which sometimes remove logos probably know how to do it.
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  26. Could you please repharse or show what you want to achieve?
    Do you want to remove the ballpen writing?
    Do you want to enhance it?
    Or do you just want to remove the noise?
    I'm totally confused in what the goal is.

    ---
    Not really knowing what the goal is I cropped left=800, right=1800, top=800, bottom=200 to limit the size (sorry, my gpu is to old and has not enough VRAM to handle more) and applied Vapoursynth(DPIR+DeSpot) on the clip.

    ---
    just noticed there ist some writing in the lower left corner of the calendar, is that what you mean?
    ... Mobile 0036 20 317 7859 ... (or something like that)
    crop:
    left: 1120 top: 1660 right 1600 bottom: 352
    Image Attached Files
    Last edited by Selur; 17th Sep 2021 at 15:54.
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  27. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by Selur View Post
    I'm totally confused in what the goal is.
    Nobody knows what the point of this meandering NeatVideo-fellating thread is about anymore (not that it ever had a point). It's not just you.

    I was only half following it to see what sort of Avisynth upscale wizardry you'll attempt.

    ... but you're getting crap for source, so not sure I care anymore.
    Quote Quote  
  28. Originally Posted by Selur View Post
    Could you please repharse or show what you want to achieve?
    Do you want to remove the ballpen writing?
    Do you want to enhance it?
    Or do you just want to remove the noise?
    I'm totally confused in what the goal is.
    I don't think English is his first language... but I think he wants to make everything more legible - he was impressed with Neat Video such as in post 54 thinking it was some "AI" - but as explained earlier - it's just motion compensation with a temporal radius. You can get similar results in avisynth/vapoursynth with a basic script
    https://forum.videohelp.com/threads/403073-Why-is-Neat-video-the-best-video-denoiser/p...e2#post2631174

    I won't upload video, because anyone can see for themselves. But it's just a basic starting point. If you wanted to you can use curves to increase legibility of text , or if you improve it farther , you can natron to stabilize a range of frames and use a median/mean filter for stacking, or even before applying denoiser. All open source. Of course you can do that with neat video too. But this is just "apples to apples" denoiser to denoiser comparison

    Code:
    SMDegrain (tr=9, thSAD=1000, refinemotion=false, mode=0, contrasharp=false, PreFilter=4, truemotion=true, plane=4, chroma=true)
    eg. frame 466

    If you compare post 54 screenshot, SMDegrain details/text are more legible, but his screenshot is scaled, it's a different frame - it' s not directly comparable. And he probably didn't use neat video correctly, maybe not ideal sampling or settings
    Not really knowing what the goal is I cropped left=800, right=1800, top=800, bottom=200 to limit the size (sorry, my gpu is to old and has not enough VRAM to handle more) and applied Vapoursynth(DPIR+DeSpot) on the clip.
    you can if you want to; vs-dpir has tiling options now . It might be useful as a prefilter . But for this type of noise , where there are "missing" pieces of signal (text) behind the noise, temporal denoising is usually more beneficial, as the "missing" pieces are covered over a few frames. Stabilizing before denoising can be beneficial too, and then reapplying the original motion if desired.

    EDIT apparently 12.5MB is the max size for PNG here, it won't let me upload the corresponding source frame. I'll zip it up
    Image Attached Thumbnails Click image for larger version

Name:	sm000466.png
Views:	19
Size:	8.38 MB
ID:	60782  

    Image Attached Files
    Last edited by poisondeathray; 17th Sep 2021 at 19:14.
    Quote Quote  
  29. Nice settngs with SMDegrain. (found a bug in Hybrid through that, I accidentally restricted tr to max 3 )
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  
  30. Using vs-dpir as prefilter for SMDegrain doesn't really help. (+ is pain since vs-dpir is RGBS only)
    Using vs-dpir after SMDegrain clears the noise a bit:

    without vs-dpir:

    all-in-all it's probably not worth spending more time on this till Truthler explains that he's aiming for.

    Cu Selur
    users currently on my ignore list: deadrats, Stears555
    Quote Quote  



Similar Threads