VideoHelp Forum
+ Reply to Thread
Page 3 of 4
FirstFirst 1 2 3 4 LastLast
Results 61 to 90 of 115
Thread
  1. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Having reviewed your denoising I do not see any use of MCTD, why ?
    Well, I have not used MCDT because I was not aware of it. The problem with the filters and plug-inīs is that there are so many of them , without a ranking about their intrinsic quality and value itīs a difficult task to choose the best ones and the best associations as it is impossible to test all of them. So I tried basically some of them that I saw more frequently referred in scripts in these forums however some of these scripts are older other are made by people with very different knowledge levels , so itīs not an easy task for a newcomer. Itīs an area where I think professionals or amateurs with a long experience could give an hand publishing a rank or at least a selection of the best and more useful usually needed filters/plug-in.

    The MCDT plug-in that you recommended seems to work fine however itīs quite a complex and extensive script with lots of dependencies and some of them needing avisynth 2.6 , I have already upgraded to 2.6 standard version not MT , and decided also to upgrade all plug-inīs associated, as I was still using version 2.5.8 ( basically because I have tested videocleaner that doesnīt seem to like version 2.6 ).I have used the complete avsi as to try to truncate the script and use only the denoise section without PP will have forced me to see it in detail and with my currently knowledge of avisynth I doubt I could have done it successfully .

    MCDT script worked fine and I think the results are good it seems to better clean media and retain more detail and so far it seems to give the best results .However itīs really extremely slow. I have changed autolevels for the first script that converts to RGB32 as I saw some comments from videoFred relating that the autolevels plug-in appears to work better in RGB. The conversion speed to RGB with the inclusion of the autolevels was not affected and the remaining restore script without the autolevels and even with the extremely aggressive settings in QTGMC as gone from 1,5 fps to something like 10-12 fps already quite acceptable. However Including the MCDT the figure goes down to something like 0,3-0,4 fps its really complicated to use it with this speed even if I modify QTGMC it probably will not give better results than 1-2 fps ( I have not changed QTGMC settings only to guarantee that the results are comparable and that any difference is not due to QTGMC)

    I tried to speed-up the process using GPU=true and including FFT3Dgpu plugin however the script says I didn’t have it or it didnīt recognize FFT3Dgpu ( I don’t remember the exact error)

    I tried also Neat video version 4 and it works quite fine, better than my previous approach however I think the detail was marginal better with MCDT, but I have not tried to optimize the noise profile or choose the better frame to do it . I used the automatic mode on the first frame an even so the results are quite good .

    And still I have not used neat video correctly as I have used it as last step in virtualdub after the script which last line was Unsharp and I think all denoise must be made before sharpening and also I do have an extra color space conversion as I was obliged to go from YV12 to RGB32 as neat video only works on RGB . So if I use neat video correctly eventually doing the sharpening only inside neat video and not using unsharp or using it before unsharp and eventually only for denoise ( I think Unsharp filter is one of my good plug-inīs) I will get probably best results namely if I optimize the neat video noise profile doing some fine tunning .Rendering speed is quite good as neat video could use all 8 cores and the GPU ( for neat video alone is something like 65 fps). Perhaps optimizing neat video itself and inserting it correctly on the script could provide better results than MCDT and rendering speed is much better.

    I was trying to avoid a scene by scene treatment and trying to find a script that could do a good job even if not excellent for one full tape as I have almost 100 tapes only from video8 ( not speaking of VHS , miniDV and super8 transfers) most of them with 90 min so almost 150 h and it will be very difficult for me to have time to do a scene by scene restore. I was only considering particular problematic scenes to be object of a different /special restore approach, however perhaps this will not be achievable but Iīm wanting to try.

    Related to HUUFYUV and UTvideo I have used the later basically because HUUFY doesnīt support YV12 and I wanted to avoid unnecessary color space conversions and uncompressed format itīs why I have used Utvideo I could use uncompressed RGB or I am obliged to change the color space , but as these are only test versions and not the final version I can use Huffy and convert from YV12 to YUV2 .

    Related to audio I don’t know if you are referring to the wav file that I tried to improve with iZotope , or the audio included with previous clips ( the last clips don’t have audio to get a smaller file). However if itīs mono I made something wrong as the original camcorder that is 26 years old (unfortunately broken ) was stereo HiFi !.... and the raw captured files have stereo audio.

    I have tried also to make 2-3 captures and use the median function to reduce noise but I have not got very interesting results perhaps these method is more useful for more problematic and noisier tapes. The noise was reduced but so was the detail so in this context it doesnīt seem very useful for these type of tapes or eventually I have not done it correctly.


    recently I was impressed by a guy ( Mattias something on Vimeo) who made a double 8mm transfer with two different apertures which brings a very impressive sensation like the contrast masking 'trick' ( it works BTW you should also consider it when it will come to editing time ; but again it has to be applied sequence by sequence not once for all )
    I dont understand this, how does this applies to my captures


    I was aware of this post to get rid or reduce the edge color issue and I have tried it and it worked . However due to the crop deshake and resize this area as somewhat been reduced and the problem is somewhat more limited ,but yes I could scene by scene try to apply this type of correction but itīs complicated due to the work involved only if I could apply it to the full tape without overall negative impacts namely on detail.

    I remarked I had in my last script used CNR2 before QTGMC however it must be used only with progressive stuff , after deinterlacing I changed that however I don’t see much of a difference.

    Even if Iīm getting better results I still have to address these non solved or not correctly solved problems:
    • Better color correction ( autolevels is not enough), I will try Msu color enhance or autoAdjust to see if I can get anything better
    • Defliker , I still need a solution for extreme luma changes with ambient luminosity ( ReduceFliker, Msu defliker ?)
    • Delogo for removing date stamping before deshaking
    ~

    If I can find relative good solutions for all of them I think the script could become a reasonable one.

    I will put samples with MCDT and Neat Video as soon as possible
    Last edited by FLP437; 23rd May 2016 at 23:19.
    Quote Quote  
  2. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    The MCTD script I mentioned earlier is slow but from my own use not as slow as the figures you mention .That’s where you wanna take time to tune the script and not rely on default settings if that’s the case. Wiki/Doom9 threads will provide extended infos for this

    On almost an ‘antique’ machine which I dedicated to Avisynth and other AV non commercial things , that is a Xeon 5460 (aka vanilla 9550), and 8GB Ram, I get >5fps with MCTD for 960*720*50 resolution (and >12fps with QTGMC ) ; So you should definitely use MT as 2.6 or 2.5.8 if you still prefer !

    I am done with evaluating these denoise scripts and so far MCTD seems a very good choice. it looks I can get rid of ‘normal’ low video noise with almost unnoticeable added blur . As for the ‘extra’ PP features associated in MCTD I may use… or not

    Now one should also remember that from using QTGMC yet most of this ‘normally-low’ video noise has been already smartly reduced without objectionable smoothing too .

    In QTGMC, considering V8/Hi8 , I do not see the burden of (trying to..) treat the noise as a separate ‘valuable aspect’ ( say like subtle film grain ) via all the heavy MatchSource/Denoise knobs. Keeping video8/Hi8 noise will not make these old videos look like vintage 70mm Cinerama … video8 noise IS bad , period

    Using the more refined SourceMatch features apply (at best) to HD sources, not video8/Hi8

    MCTD should be then applied only on scenes which still feature noticeable remaining noise ; And yes unfortunately this will have to be tweaked for each scene (via Sigma at the very very minimum) . That’s not quite confortable for large projects . Indeed . And I still have not made up my mind on that or go NeatVideo

    denoise should be made right at the ‘beginning’ just after deinterlace ; and so alas on a scene by scene method . Idem for possible dehalo, antialias . Now I don’t’ have so many scenes I intend to keep and that would require these prior steps (before going to the timeline)

    >> I dont understand this, how does this applies to my captures

    well I simply mentioned this guy which achieved (yet 2 years ago) some HDR on 8mm film using two exposure time ; smart ! this is quite similar to hereafter and it proved great as well the way he did

    That is you may take interest in ‘contrast masking’ once you will get your footage in the timeline. While I am still reluctant in any kind of automatic AGC for an expected HDR bonus, you can do a much better job ‘manually’ using the well known HDR ‘trick’ ( duplicate layer / desaturate / invert /choose adhoc radius for Gaussian blur/ choose adhoc setting % for overlay ) ; that works , while auto AGC hardly will and almost imprevisibly too ; indeed you then have the ‘symetric’ alternative layer pratcice to darken over -highlited areas , just very useful too.

    As for the point of upsizing D1 low resolution , so far I have not found any SR that works with my video material . I tried SuperRes and was not impressed even in the best case there is ( size 2) . The principle for it is (unless I am wrong) resize such that 1=4 pixel that take into account previous and next frame infos for this new size2 pixel. It looks exciting at first glance for video but in the end I found using SuperRes merely alters very slightly (up) the britness which some may then associate as…’better’ or ‘sharper’ ! while after careful realign I was hard pressed to then find any size2 SR new video worth the name “super resolution”

    Sound is normally ‘no problem’ and I keep it as dubbed separate tracks for extended sound control outside Premiere. Having control at the ms level enables often very easy replacements ( copy/paste) instead of painful still audible ‘cleaning’ of, for instance, sudden wind blow burst in the mic ; etc..

    Considering the point of D8 vs ‘any player’ replay/capture , it simply may be that EV9000 playback is definitely no different from the better main street camcorders of the time ! As a rule Sony was careful not to bring “too performing” features in their prosumer Hi8 *and* prosumer DV range ….

    I expected a 3ccd V6000 but Sony released a shaky handheld ( no stab!) 3ccd VX1 . Never mind as the V6000 itself was poorely designed ..by design (!) as a shoulder camcorder. I felt uncomfortable too with it almost having to use it as a much too heavy handheld device not a true shoulder device . Crapules


    BTW how does it happen that capturing via EV9000 , some 6/8 or so bottom lines are missing !!?? which as a side effect ended here as some appreciation errors too

    D8 looks the best solution if one forgets (forgive) the infamous right side color inversion(s) ! especially since this nasty thing only applies to PAL D8 versions , NTSC is free of these artefacts

    ps : no more christmas tree smileys ...I make progress
    Last edited by kerryann; 24th May 2016 at 11:05.
    Quote Quote  
  3. QTGMC and MCTD together are very slow. If you're using QTGMC you might as well try its denoising instead of adding another denoiser. Something like:

    Code:
    QTGMC(preset="slow", EZDenoise=1.0, DenoiseMC=true)
    There are many variables you can adjust.
    Quote Quote  
  4. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    Hi jagabo, Indeed EZDoise is quite performing. Still you won't (afaik) get the very same inner level of fine noise control as MCTD.

    edit : I also think it is slow since he does not seem to use MT ...
    Quote Quote  
  5. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Thank you jagabo I will also try QTGMC denoising and between the three options MCTD, QTGMC and Neatvideo I will try to find the best trade-off between quality and speed . I have no special problem with slowly scripts I prefer quality to speed , however I still have some limits and any thing that takes more then 12-14 h for rendering a 1 hour tape ( more or less 1,5-2 fps ) begins to be a little to much for me.

    That said I have probably to try avisynth MT is quite likely that it can speed-up the script . My only problem is that I have a lot of products that I have been testing and each time I change the avisynth version there is one that stops working or oblige me to see in detail all the environment.

    Iīm posting the first result with MCTD I think the results are better than any of my previous attempts. I also made some small changes as the autolevels filter seems to work better in RGB , and also give better results removing black borders before applying it so I changed the first script to convert to RGB to

    Code:
    AviSource("C:\name.avi")
    ConvertToRGB32(matrix="PC.601",interlaced=true)
    Crop(12, 4, -12, -8)
    autolevels(autolevel=true, autogamma=false)
    After in VirtualDub I just do the deshaking ( crop and resize just before and after deshaking have been changed.Crop as been transposed to this first script and Iīm doing only the final resize at the end of the restore script).I also changed the resizer from LanczosResize to Spline64Resize hoping that it will be more precise. I still intend to have only one script for all the workflow I have no doubt that is possible is only a question of time , and I am more interested now to try to identify the filters, settings and correct relative positioning. I will compact everything when I have a stabilized workflow.I have not changed QTGMC parameters only to maintain the results comparable with previous results.

    restore script

    Code:
    AviSource("E:\Video8\deshake.avi")
    AssumeTFF
    QTGMC( Preset="Very Slow", SourceMatch=3, Sharpness=0.5, TR2=2, Lossless=2, MatchEnhance=0.75, NoiseProcess=1, NoiseRestore=0.7, Sigma=1.5)
    santiag(strh=2,strv=2)
    MergeChroma(aWarpSharp(depth=10), aWarpSharp(depth=20))
    Overlay(last, ColorYUV(off_y=-2, off_u=5, off_v=-4), 0, 0, GreyScale(last).ColorYUV(cont_y=30))
    Cnr2("xoo",4,2,64)
    ChromaShift(C=-2, L=-2) 
    MCTemporalDenoise(settings="low", thSAD=256, thSCD1=256, radius=2 , sigma=5 , bwbh=32, owoh=4, blksize=8, overlap=4, maxr=4, TTstr=3, adapt=true, protect=true, enhance=false, truemotion=true, stabilize=true, chroma=true, gpu=false)
    UnsharpHQ(THRESHOLD=20, SHARPSTR=5.0, SMOOTH=0.5, SHOW=false)
    Spline64Resize(720, 576)
    Iīm still searching for solutions for defliker, color correction and delogo, the ones that I tried defliker, reducefliker and autoadjust didnīt gave significative better results, but sometimes I donīt know if are the filters themselves or incorrect parameters selected by me.

    ps-the video as still been obtained with LanczosResize in the script, however I donīt think that any eventually difference from Spline64Resize if any, will be visible.I provided the video in Huufyuv, witch adds one additional color space conversion but it seems to be easier for kerryann.
    Image Attached Files
    Last edited by FLP437; 25th May 2016 at 22:43.
    Quote Quote  
  6. restore.6.avi is way oversharpened in my opinion. There are halos around every sharp edge.
    Quote Quote  
  7. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Yes you are right I will try to reduce sharpening parameters on QTGMC or/ and in UnSharpHQ
    Quote Quote  
  8. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    too much sharpen , indeed.

    0.5 is normally OK in QTGMC especially with TR2=2 ; so yes look in UnSharpHQ first .

    I would be interested to see the very same sequence capture as page2 "raw_D8" Huffyuv but as interlaced DV.avi , and indeed as straight transcode through your D8. This with Chroma filter On and miniTBC as well
    Quote Quote  
  9. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    I tried two approaches first reducing the sharpening ( UnSharpHD only or both UnsharpHD and QTGMC) second reduce the halos without reducing sharpening using the DeHalo filter immediately after UnsharpHD.

    Iīm including 4 frames pictures

    1- Frame from restore 6 no modifications to compare purposes only
    2- Applied Dehalo filter ( DeHalo_alpha(rx=4, ry=1))
    3- Sharpening reduced in UnsharpHD ( THRESHOLD=25, SHARPSTR=4.0)
    4- idem as 3 but changed also QTGMC sharpeness from 0,5 to 0,2

    Iīm not sure but I think the best seems to be the result from the DeHalo filter, I think it provides better results than reducing the sharpening but I trust more in your opinions than mine as sometimes I donīt detect important details, only after someone points them to me.

    I would be interested to see the very same sequence capture as page2 "raw_D8" Huffyuv but as interlaced DV.avi , and indeed as straight transcode through your D8. This with Chroma filter On and miniTBC as well
    I can do that but perhaps you can first see two other thread on the capture section where I have discussed somewhat these problems and if you donīt find what you want , I will do it for you.

    . Conventional analog capture workflow vs alternative methods - comparison
    and eventually also
    . Startech USB3HDCAP opinions

    The conclusion related to the D8 Cam were that the DNR did nothing and sometimes did worst than nothing, and the TBC was very limited. In my current capture configuration using the Panasonic as an intermediate ADC and TBC it was best to rely on the Panasonic TBC like features than on the line TBC from the D8 camcorder . Using the D8 internal ADC and Transcoding directy to DV with the D8 the line TBC included is better then nothing and was still useful using.


    Related to the DV vs Svideo capture the difference was not enormous and even some members seemed to prefer the direct capture via DV

    As I have told when I have a little more spare time I want also to try the card that I bought some time ago the Magewell Pro capture HDMI using the direct svideo capture ( as it captures analog video to ) to see if it could be better than the internal Panasonic ADC as it has a pipeline with four 12 bits ADCīs and what they call an , Advanced time-base correction (TBC) with frame synchronization included , it also does deinterlacing (hardware based) .However all of this depends on how good is the implementation done around the the 2 mains chips by Magewell only after trying I will know. I want also to try but in this case I donīt expect any substantial difference one capture using the HDMI input using the same workflow I have today ( Panasonic as intermediary ADC and TBC outputting to 576i HDMI).

    Related to MCTD is it possible to use also the GPU ? if possible it could speed up the script, I tried gpu=true and even if I have the FFT3Dgpu in the plugin folder it gave me this error
    Image Attached Thumbnails Click image for larger version

Name:	avisynth error.jpg
Views:	369
Size:	30.1 KB
ID:	37153  

    Image Attached Files
    Last edited by FLP437; 25th May 2016 at 22:14.
    Quote Quote  
  10. Originally Posted by FLP437 View Post
    I tried gpu=true and even if I have the FFT3Dgpu in the plugin folder it gave me this error
    Do you have the FFTW3.DLL where it's supposed to be? Depends on the operating system. In my 32bit Win7 it's in the System32 folder.

    Iff you're using QTGMC in your script, why not have it use one of its denoisers? DFTTest is very good.
    Quote Quote  
  11. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Well, theoretically yes . I have a windows 7 , 64 bits so it is on the C:\Windows\SysWOW64

    However the error says no function name FFT3Dgpu it seems related to a not identified function on the script , but itīs strange as gpu is one standard variable ( by default is false but can be true ) and the script with the gpu=false works fine

    If you're using QTGMC in your script, why not have it use one of its denoisers? DFTTest is very good.
    I have made preliminary tests with EZdenoise and Neat video (external filter) and both work also fine I only need more in depth tests to decide which one to retain between MCTD, QTGMC denoisers and Neat Video, I have not tried yet DFTTest I will include it in my tests also.
    Quote Quote  
  12. Oh, GPU=False works okay? Then you're sure you have a compatible video card? Not every card works.
    Quote Quote  
  13. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    Originally Posted by FLP437 View Post

    I can do that but perhaps you can first see two other thread on the capture section where I have discussed somewhat these problems and if you donīt find what you want , I will do it for you.
    yes, please do if you can . I would yet simply like to see what comes out as straight D8 interlaced DV.avi versus the all Panasonic setting you use.

    D8 vs 'the rest' has also been long and hard discussed in french video forums since a decade
    Quote Quote  
  14. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Ok I will do this capture but perhaps it will be better based on what I have learned do it with Dnr=off and Tbc=on as it seems to be the settings that provide better results , because apparently dnr doesn't seem to provide good results and also because there is a tendency to capture with noise as denoise avisynth filters could be more efficient.

    And do you prefer the first clip I used on page 2 or the last clip I have been using
    Last edited by FLP437; 26th May 2016 at 09:58.
    Quote Quote  
  15. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    oki , do something that matches this "D8_raw" huffyuv interlaced file capture then

    Since on the advice of the chroma NR filter 'should be off' as it is best to capture chroma noise and give it to Avisynth...well no comment.

    inside their camcorders Sony definitely used of appropriate filtering . It was even said at the time of the V6000 camcorder that Sony would almost regret the presence a 'too good chroma' filter for a prosumer device ! (> 45dB on chroma alone)

    I found chroma ON to beOK...lets say at least on my tapes then ; with little color bleeding ( for Hi8 indeed !)

    ps: since it is very short duration , let's see the two versions on/off then !
    Last edited by kerryann; 26th May 2016 at 11:37.
    Quote Quote  
  16. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    Originally Posted by FLP437 View Post
    And do you prefer the first clip I used on page 2 or the last clip I have been using
    you asked about your 4 pictures. I just had a look (been cruising for gas for 2 days ) and at first glance I would say :

    1) rather sharp such as according to your script ; some halos are noticeable ... on this static picture; it remains to see if they would still obviously show once in H264 format (loop pp, etc..) and viewed as a 50fps movie from a normal distance

    2) Dehalo...why not but the script used here is possibly not well tuned enough imho ( if there are that many 'knobs' for it , I have not yet really looked at this , I intend to do it in a near future) ; as proposed here the filtering does not improve picture overall imho (blur to start with)

    3 and 4) I 'would favor' these given the 2) point , with mixed feelings though
    Quote Quote  
  17. For dehalo try:

    Code:
    Dehalo_alpha(rx=3.8, ry=3.4).Sharpen(0.3)
    Unfortunately, dehalo_alpha kills some detail. It's best not to generate them in the first place. But that's pretty hard to do because most decks apply a sharpening filter that creates them.
    Quote Quote  
  18. You always want to turn off DNR when capturing. The filtering circuits are crude, even in the best VCR, and do terrible things to the video. If you want to do filtering or sharpening, do that in software. TBC should be on, but you should always experiment because some TBC circuits actually make some video worse. This is not true of professional TBC, but most TBC circuits built into prosumer VCRs are not professional.

    Bottom line: you cannot find a "recipe" and then use it on every tape. You always need to take a few minutes and check your settings before proceeding with the capture.
    Quote Quote  
  19. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Here there are the files requested by kerryann

    1 - D8_DV_raw - direct DV capture from the D8 Cam. (DNR=off, TBC=on)
    2 - D8_DV_restore_Old - the DV file "restored" with the old script as per page 2 identically to the one used on D8_raw.avi only for comparison purposes
    3 - D8_DV_restore_New - The DV file "restored" with the last script version ( including already the new settings for Dehalo_alpha as recommended by jagabo
    4 - D8_restore_last - The raw file captured through svideo with the workflow including the Panasonic dvd recorder and the hdmi capture card "restored" also with the last version of the script including already the jagabo DeHalo improvement.

    I think the new settings proposed by jagabo for the Dehalo have resulted well so far I think these are the best results , thank you so much

    However I still have to finish the denoisers tests to see which one could give the best results , in preliminary tests ( with settings not optimized ) all of them have resulted well however there are enormous disparities in speed between MCTD the slower of all of them and Neat Video the faster (measured alone and not optimized) , If I choose MCTD I will have to find a way to speed it up ( avisynth MT, GPU use , whatever) if I have trouble to decide because they are all very similar probably I will choose neat video or one of QTGMC denoisers as they are faster.

    I still have to address the defliker (rapid luma changes with over and under exposures due I think to the Cam AGC), and delogo ( clean time stamping before deshaker ) problems and try to find solutions , also better color correction if possible. However I think the autolevels is now working a little better.
    Image Attached Files
    Last edited by FLP437; 27th May 2016 at 20:48.
    Quote Quote  
  20. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    thanks for the re-capture. Too bad you did not provide the ChromaNoise filter 'ON' as well as a second file versus the 'Off' version.
    It might be it was better ?

    I have no ego point here , I simply (as a personal rule) want to stick to plain simple facts . Since -they- rule, don't they ?

    So what I did is the very usual basic procedure : open side by side in VDub the two files and then being interlaced footage, I looked for some 'clean' static pictures as they are simpler to deal with versus moving fuzzy blurred pictures which can be misleading to appreciate .

    There are some , hereafter is screen capture of frame 35 , there are some others around 60+ , and a few others here and there.

    first point , and I would say as usual , one has to be able to deal with a slight brightness difference here which is then a very common pitfall where many fall stating "ha..look it's better" ; now almost by chance it remains rather reasonable difference and then easy to deal with ...for an agnostic ready eye ( one that has not made its mind even before having a look at facts )

    What would you say here ? is straight DV inferior or..better than the lossless capture via added Panasonic gear ? could it also be that that going through Panasonic outside 'treatment' yet apply some unwanted mild sharpen in the first place , ending with some more objectionable interlace jaggies for instance ... and it is a fact that too often people will favor sharpen (!?)

    Adding that my main very first concern was indeed that DV compression @25Mbps could bring some minor but yet visible issues @5:1 being such a close cousin of MJPEG ( well DV is not exactly plain vanilla MJPEG copy with some little clever bonus , ie some motion time prediction , but nothing that far superior anyway)

    So it is what I was interested to look for having made the final bet ( for my own captures) it would not improve other than stay simple straight D8 transcode even as DV compression . In the past I did advocate the opposite option , that 'at first glance' I would be keen to think DV transcode would "fail" versus more sophisticated procedure .

    BUT there was ( years ago yet) some tough discussions in french video forums with people having caustiously undertake this comparison with a quite impressive equipment suite ; including a'modern'(digital era) TBC worth the true pro grade pédigré (such as fine chroma handling for once) . And in the end these very same people would advise to keep...straight D8 ! Heck... agnostic eye did I say

    Again I am here in no egotic discussion , simply as looking for facts not being written beforehand.

    I mean I suspect that some who say a chromafilter should "always be turned off " may only have simple 'sharpen knob' experience with antique VHS gears , possibly ; then yes in such case they are correct . Now have they owned and used extensively a good D8 ( which one then ) and end with such 'definitive' constatation . Possibly something was broken here , I mean the D8 indeed



    @FLP437 what about a 'ON' version ? I will have a look at other (restore) pictures latter
    Last edited by kerryann; 28th May 2016 at 08:20.
    Quote Quote  
  21. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Ok I will post a dnr=on recapture
    Quote Quote  
  22. Originally Posted by kerryann View Post
    I mean I suspect that some who say a chromafilter should "always be turned off " may only have simple 'sharpen knob' experience with antique VHS gears , possibly ; then yes in such case they are correct . Now have they owned and used extensively a good D8 ( which one then ) and end with such 'definitive' constatation . Possibly something was broken here , I mean the D8 indeed
    I have lots of experience, and the NR filters in all consumer and prosumer VHS decks are extremely crude and do nothing but destroy detail. Most of them are simple low pass filters.

    The last VHS VCRs were designed in the early 1990s. At that time, all the circuitry in the signal path was analog. The signal processing afforded by these circuits is absolutely stone age compared to what can be done with even the simplest AVISynth script.

    If you doubt my advice, just look at your own captures.
    Quote Quote  
  23. There's something very odd in D8_DV_restore_New.avi. The motion of the background is smooth but the motion of the man is herky jerky. Like the field order on the background was right but wrong for the man. I suspect the field order was set wrong before QTGMC() and the video was later run through a deshaker which smoothed out the panning.
    Last edited by jagabo; 28th May 2016 at 12:16.
    Quote Quote  
  24. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    Originally Posted by johnmeyer View Post
    I have lots of experience, and the NR filters in all consumer and prosumer VHS decks are extremely crude and do nothing but destroy detail. Most of them are simple low pass filters.

    The last VHS VCRs were designed in the early 1990s. At that time, all the circuitry in the signal path was analog. The signal processing afforded by these circuits is absolutely stone age compared to what can be done with even the simplest AVISynth script.

    If you doubt my advice, just look at your own captures.
    well, with lots of experience what would you expect else than a ... "simple low pass filter" for limiting chroma ( frequency bleed ) ? some high pass filter ...

    Now the built-in comb filter in a Sony D8 is a bit quite more elaborate than what you think

    If you refer to your own old VHS vcrs , then okay , no pb

    Otherwise if you refer to any specific D8 you may have owned and used (?) and found the CNR was "extremely crude and do nothing but destroy detail" please mention the one you have used and prove this --with facts ( demo files)
    Quote Quote  
  25. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Thanks jagabo, you do have linceīs eyes, I have not remarked anything. Well this is really weird, I thought I forgot to make an explicit assumption in the first script that the file was BFF, and I thought it have assumed TFF. But having done so and also assuming BFF in the second script after deshaker is the background that becomes jerky. I have to assume TFF in the second script to have a smooth background movement but the man remains with a jerky movement. Something is happening just before or after deshaking I donīt know what could be causing this as the capture through svideo doesnīt have this problem , could it be something related to the dv capture itself ? it a first time never happened before to me.

    However I think kerryann is more interested in the raw files perhaps no need to recapture again or try to find what is causing this situation.

    @kerryann
    Related to my capture workflow I saw some month ago some minor references to this method on digitalfaq forum, after I found a german forum where this type of workflow was being discussed in detail

    http://forum.gleitz.info/showthread.php?46713-Zeitgem%E4%DFes-hochwertiges-analoges-Ca...er-HDMI/page55

    The problems I had before never completely well solved by me were capture quality and stability and I made a painful journey trying more than a dozen perhaps more capture cards and several players some new, some old, I begun with the cheaper ones and evolved progressively (startech usb3hdcap (svideo), osprey 210e, ATI HD600 usb, hauppage usb-Live 2, Terratec G3, startech usb2HDcap, several old devices that I had Philips SAA7134, dazzle dvc100 ,pinnacle pctv usb, grass valley advc 55, pinnacle moviebox 710 usb, etc ). Even if some definitely have better image than others I had always a problem with analog captures through svideo , lost frames ,more or less depending on the card ( one exception as been the osprey card that as been the only one able to do captures almost without lost frames ). So I was obliged to use an external TBC, I used the AVT 8710. And from my point of view the results were not good,apparent loss of detail and I discovered that it raised the black levels and at the same time lowered the white levels, greatly reduced horizontal resolution and shifted the image a bit to the left. So I begun trying using DVD recorders as TBC like devices. I had a LG dvd recorder and as results were better, I continued this path after with the Panasonic which has provided even better results.

    However captures using DV using inicially the advc 55 or the moviebox 710 and after through the D8 camcorder usually not led to loss of frames.Even VHS direct capture in DV with my JVC deck that have DV output also worked fine without lost frames. However the quality seemed always not so good as from svideo.

    What are the main advantages for me from this workflow. I never had to worry again about lost frames ( Panasonic TBC like circuitry works well) , capture quality was equal or better to any of the capture cards I have tried ( I have still to try the magewell pro hdmi analog input), capture device's Proc Amp adjust automatically in 8-bit on the Pana and capturing in HDMI I never more had to worry with levels adjusting in the processor amplifier as there is no possibility of grabbing some values that would be clipped above 255 or below 0. So more or less a clean capture without the habitual hassles.
    And I almost forgot the removal of one additional ADC cycle due to the external TBC
    Last edited by FLP437; 28th May 2016 at 20:09.
    Quote Quote  
  26. Originally Posted by kerryann View Post
    well, with lots of experience what would you expect else than a ... "simple low pass filter" for limiting chroma ( frequency bleed ) ? some high pass filter ...

    Now the built-in comb filter in a Sony D8 is a bit quite more elaborate than what you think

    If you refer to your own old VHS vcrs , then okay , no pb

    Otherwise if you refer to any specific D8 you may have owned and used (?) and found the CNR was "extremely crude and do nothing but destroy detail" please mention the one you have used and prove this --with facts ( demo files)
    Your post makes it appear that you do not understand what low pass filter does to video, and how it would be used. You then mention a high pass filter, but I have no idea what point you were trying to make.

    Also, you mention a comb filter, but you clearly do not understand what it does. I say this because a comb filter has nothing whatsoever to do with noise reduction. It's purpose, when used with analog video, is to provide a cleaner separation between luma and chroma subcarriers. If properly implemented, it can reduce dot crawl. Dot crawl has not been mentioned in this thread, and has nothing to do with the OP's issues. Therefore, mentioning a comb filter is both irrelevant and misleading.

    You want me to post files to prove my point that hardware-based noise reduction in these old VCRs and cameras is vastly inferior to the noise reduction that can be done with AVISynth plugins (or Neat, etc.), but I'm certainly not going to take the time to do that. I'm sure you will then claim that this proves I am wrong (you've done this before), but that is silly. So, rather than waste my time posting my own files, I once again make the same statement I made before: the OP's own captures will show that the noise reduction circuit will make the video very soft, with significant loss of details, especially when compared to capturing without noise reduction, and then doing noise reduction in software.

    This is true whether the deck is a D8, Hi8, 8mm, Beta, VHS, or S-VHS. What we are talking about here is the analog noise that is common in any analog source from 2-inch Quadruplex down to 8mm tape.

    In answer to your question about my current decks, while I don't own a Digital 8 deck, I do have 8mm and Beta decks, as well as the following S-VHS VCRs:

    JVC VCR SR-VS30U (very rare DV / S-VHS deck, with built-in capture and Firewire output)
    JVC VCR HR-S4500U (S-VHS deck, with jog/shuttle and insert editing capability)
    Panasonic PV-S4990 (one of the best prosumer S-VHS Panasonic decks, although not as rugged as the true pro decks)

    I also have a really cheap Sharp VCR that I use mostly for rewinding, not so much to avoid wear and tear on the better VCRs, but because it has a super high speed transport.

    I'm trying to help the OP by making sure he doesn't degrade his end result by capturing tapes with the noise reduction turned on. Having said that, I do acknowledge that features on different VCRs often have confusing names. For instance, the TBC circuit sometimes has a switch or menu setting that is labeled "noise reduction." Obviously a good TBC is essential when doing quality captures from analog tape, and if you need to flip a switch labeled "noise reduction" in order to get time-base correction, then by all means turn that switch on.
    Quote Quote  
  27. Originally Posted by FLP437 View Post
    The problems I had before never completely well solved by me were capture quality and stability and I made a painful journey trying more than a dozen perhaps more capture cards and several players some new, some old ...

    However captures using DV using inicially the advc 55 or the moviebox 710 and after through the D8 camcorder usually not led to loss of frames.Even VHS direct capture in DV with my JVC deck that have DV output also worked fine without lost frames. However the quality seemed always not so good as from svideo.
    I have only quoted part of your post, so it may be a little misleading, but what I think you are saying is that you got much more repeatable and reliable captures when using DV capture. This has certainly been my experience, and even though lots of people here don't like DV capture because of the 4:1:1 (or 4:2:0 for PAL) color space, for consumer analog video, you've got a lot more serious quality issues to deal with, and the 4:1:1 isn't going to show up much. This is especially true for NTSC where the colors towards the red end of the spectrum are going to be completely wrong, not matter what (according to some wags, NTSC = Never The Same Color).

    So, if you can't get your capture card to work reliably, stay with DV, and don't worry about it.

    A decade ago, when digitizing analog tapes was a far more common practice, I wrote quite a bit about capture, and often asked the question, "how many dropped frames is allowable?" My answer: none. Unless the tape is damaged, you should never have even one dropped frame.
    Quote Quote  
  28. Member
    Join Date
    Mar 2015
    Location
    Europe
    Search Comp PM
    Thanks johnmeyer, what you say is true and itīs also my opinion, itīs easier to get more repeatable and reliable captures when using DV , but now with my actual capture workflow I think I do have a stable and good quality capture process .However I will have always an open mind towards improvements if they do happen .

    On the other hand I have also almost completed my tapes collection capture process ( I have done this several times as my workflow as improved over time ). So on the capture side if no disruptive improvement happens ,Iīm quite comfortable now, this discussion about DV / Svideo DNR=off / on , TBC, etc I already done it some months ago.

    However in the restoration side Iīm still slowly and grievously trying to improve my restoration workflow and scripts and now is in this area that Iīm more in need of support in order to try to finish a generic restoration script that could be used for video8 and vhs and could serve as a base that I could after adapt for specific problems I may find.

    If in this area you do have any advice's or a good method for solving problems still not yet correctly addressed by me, as cleaning time stamping before deshaking , solve the sudden over and under exposure due to, I think the camera AGC or also a better solution than autolevels to adjust color and luma you are much welcomed as they are the major problems that I feel Iīm facing now.
    Last edited by FLP437; 28th May 2016 at 19:28.
    Quote Quote  
  29. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    @johnmeyer


    well, John, do you always ridiculously "twist" other people messages up to alter totally what THEY said ? what are you after doing this?


    you quite 'brillantly' copy/paste very very basic things everyone there knows (!!) , even if they sometimes hardly address the point in discussion.

    who do you want here to teach what any type of filter is made for ! such as impressively quoting a comb filter can provide improved filtering for luma/chroma separation ; who do you think to fool with these basics ? is that your "lots of experience"

    but anyway...since you also seem to have some remaining difficulty with these things , let me friendly give you one old and simple link ( in English) related to this matter of *comb filter* applied to separation between luma and chroma : "understanding video comb filters" ( the very first one that came ! as indeed there are hundred others for your interest )

    http://www.broadcaststore.com/pdf/model/793698/TT201%20-%204918.pdf

    Let me also add that you make a plain first mistake speaking of "separation between luma and chroma subcarriers" ...there is NO luma "subcarrier" per se !! with such "lots of experience" you still misunderstand some basic fundamentals ; there is simply a luma MAIN carrier not a subcarrier

    and what you deny (from misunderstanting all this) hard comb-filtering the "residuals" of chroma crosstalk ( the proper word is cross luminance) still present in the luma bandwidth is plain chroma (noise) reduction ;

    and BTW I have never so far used of your imprecise "noise reduction" AIO vague term . I focus since several messages on "chroma noise" filtering. and it is achieved via comb filtering whether you understand it or not !

    And in the end you admit ....not having experience with a D8 ! nore D8 transcode !! which does not stop you from advising to by-pass its CNR since "it will ruin the video"

    truely speaking you make me waste my valuable time ; from the start I noticed your pedantic propancy to shine with ...basics ! which I would not figure you even hardly really are familiar with !

    frankly when someone dare to say "I have lots of experience" I would expect a higher level of discussion . Could it simply be you have some tremendous ego problem ? then I am sorry for you since you may be positive as well sometimes


    EOT

    ps: you almost 'stupidly' also come hard on "stone age" analog video technology. Why not ? but not from you now ! I am no name droper as you seem to be but regarding analog technology I once (almost 30 years ago) attended a private demonstration of analog TVHD known as D2 MAC and you know what ...it was brilliant tech which came too late indeed . And I also remember 819 TVL B/W (ota) we had and to see some Hitchcok (sorry for the missing 'c' !) it was brilliant analog stone age tech ! think 819vertical TVL in the 60's , not bad
    Last edited by kerryann; 28th May 2016 at 19:38.
    Quote Quote  
  30. Member
    Join Date
    Jan 2016
    Location
    France
    Search PM
    Originally Posted by FLP437 View Post
    .../...
    However in the restoration side Iīm still slowly and grievously trying to improve my restoration workflow and scripts and now is in this area that Iīm more in need of support in order to try to finish a generic restoration script that could be used for video8 and vhs and could serve as a base that I could after adapt for specific problems I may find.
    ../...
    quite frankly you start with quite good ( for video8 indeed) shots if they are all the same kind . let's not mention the shakiness here. I have not yet looked after your F500E but I would figure it was some close camcorder to the more costly V200 ; yet 480k photosites , not bad for a >25 years old prosumer.

    It was admitted that optics/lenses were not as good as Canon that's why I choosed Canon instead of a V100 , this almost 30 years ago , tempus fugit

    So from what you posted so far you should achieve some quite decent video . The capture workflow is no big deal ( siginificant difference) with what you may do on/with these pictures. The point will possibly be to ...keep it simple ! my 2 cents
    Last edited by kerryann; 28th May 2016 at 18:50.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!