VideoHelp Forum
+ Reply to Thread
Results 1 to 16 of 16
Thread
  1. Hi,

    I am having an application created which allows users to view a sports video and capture frames. I would like to have the frame grabs as high resolution as possible for printing 5x7's and possibly 8x10's.

    My question is, what format would be best? The application has been started and is being developed in flash. I heard that with Hi Def we should be able to get almost a 2MB file? but the samples I have seen (using .flv video files) are all smaller sizes.

    What would the recommendations be to grab hi res frames/photo files?

    thanks

    john
    Quote Quote  
  2. Best format would be a lossless format e.g. PNG, BMP, TIFF

    BMP and TIFF are uncompressed, so the filesize will be larger than PNG which uses lossless compression
    Quote Quote  
  3. Thanks, but are you referring to the photo being saved from the frame?

    I am questioning which video format. Can you gegt a HD (1.pMB approx.) frame grab from an FLV video file?

    thanks

    john
    Quote Quote  
  4. Originally Posted by johnp12 View Post
    Thanks, but are you referring to the photo being saved from the frame?
    Yes


    I am questioning which video format. Can you gegt a HD (1.pMB approx.) frame grab from an FLV video file?
    Then it depends on what is available and the specifics of that video. Just because something is "FLV" or "MP4" doesn't necessarily indicate anything about the quality or resolution (dimensions)


    e.g. FLV can be HD 1920x1080 or standard definition . If it was HD , yes you can get an HD frame grab. If it was SD, you couldn't.
    Quote Quote  
  5. Maybe you should provide more information.

    What is the source of the videos ? Are they user submitted, or are you providing it ?


    FLV is just a "container" format. It doesn't necessarily indicate anything about the video compression used, or the dimensions . The other commonly used streaming flash format is MP4 - and it too is a container.


    Originally Posted by johnp12 View Post
    I am questioning which video format. Can you gegt a HD (1.pMB approx.) frame grab from an FLV video file?
    Also note the frame grab "size" - as in storage space (MB) - is meaningless without specifying the compression or format of the still image used

    e.g. Uncompressed BMP 1920x1080 will give you 5.93MB, but if you used PNG, it will vary depending on the content. It might be 10kB, it might be 1MB for the same image quality
    Quote Quote  
  6. Thanks

    The original video I created with Nikon D800 at setting 1920x1080 and is in .mov format.

    I need to convert this to an FLV file. I have converted it using sothink video convertor
    but each frame grab from the flv file is only about 2ok. As you can tell I am totally new to this.

    I tried using all the largest output settings

    I also used the original .mov file in windows live movie maker and the frame grab in ong format (that is the only format it allowed me to select) was 123k

    I am sure I am doing something simple incorrectly.


    thanks again for your help
    Quote Quote  
  7. Why do you need to convert it to FLV ?

    Unless you just re-wrap the MOV to FLV, there will be quality loss when you re-encode it .

    Again, the file size in kb of the screenshot isn't necessarily an indicator of quality . Even between PNG, there are different compression levels. So the same image quality might be larger in filesize or smaller depending on the PNG compression level used

    The highest quality will be obtained from taking the screenshot directly from the source video, using lossless compression (e.g. PNG). Do NOT use something lossy , like JPEG. You could use VLC to take screenshots from the MOV
    Quote Quote  
  8. Hi,

    Why do you need to convert it to FLV ?
    The developer developing the software for me is writing it in flash, so it needs to read flv format.
    I am just learning, but I am wondering can a HD file be taken from flash or should my application be
    written in some other language that can use other video formats? I had read that an HD video should be able
    to create approx. 1.8MB files?



    Unless you just re-wrap the MOV to FLV, there will be quality loss when you re-encode it .
    what method/software can be used to just re wrap it?


    You can view the application what I am talking about here:
    http://173.13.65.186/ap_video/index.html
    login with test user johnp12 and password tester1
    once logged in just click the search button on bottom left
    then click "new event" from the result list on the right
    and then click "event 1" on the right
    then click a video


    I guess at this point I am trying to learn/figure out, was flash the wrong way to go with this project or can we get larger files from it when the videos are set up properly? Or would other formats (other than FLV) profice better results.
    I would like to be able to print out ok quality 5x7s from the frame grabs.

    Currently it grabs the frame from the actual video you are viewing. I also had a thought that if the viewing had to be in flash, maybe the user can view the flv video but when frame grabs are taken, take them from a secondary video file in a different formatted video file (That is if the best quality hi res frame grab cannot be taken from a flv file)



    thank You
    John
    Quote Quote  
  9. Originally Posted by johnp12 View Post
    Hi,

    Why do you need to convert it to FLV ?
    The developer developing the software for me is writing it in flash, so it needs to read flv format.
    I am just learning, but I am wondering can a HD file be taken from flash or should my application be
    written in some other language that can use other video formats? I had read that an HD video should be able
    to create approx. 1.8MB files?
    Adobe Flash can use MP4 or FLV for streaming video . Uncompressed 1920x1080 (ie. BMP) will give you 5.93MB files




    Unless you just re-wrap the MOV to FLV, there will be quality loss when you re-encode it .
    what method/software can be used to just re wrap it?



    ffmpeg (command line application), but "sowt" audio is incompatible in FLV or MP4, so audio will have to be converted . (I'm assuming you're talking about the D800 only)





    I guess at this point I am trying to learn/figure out, was flash the wrong way to go with this project or can we get larger files from it when the videos are set up properly? Or would other formats (other than FLV) profice better results.
    I would like to be able to print out ok quality 5x7s from the frame grabs.

    Currently it grabs the frame from the actual video you are viewing. I also had a thought that if the viewing had to be in flash, maybe the user can view the flv video but when frame grabs are taken, take them from a secondary video file in a different formatted video file (That is if the best quality hi res frame grab cannot be taken from a flv file)

    You need to provide more background information

    I'm still unclear where the videos are sourced from ?

    Who are the potential users ? Soccer moms ? Why is this being developed ?

    e.g. are these user uploaded videos ? from a variety of cameras or only D800's ?



    Without question, the highest quality and "best" format to take the screenshots from would be the original video, but not all original formats can be streamed through that flash application (only FLV and MP4, and only some types of compression) .
    Quote Quote  
  10. Hi,

    Ok, here is more background. I currently take and sell still photos at sporting events.
    We are often asked about offering video as well. My thought was to combine the 2. Offer both.
    Even though I know the frame grab quality will not be as good as a DSLR photo, but honestly,
    most people only are interested in 5x7 or 8x10's. So higher quality is overkill. and over time I am guessing
    video quality will only improve.

    I am also thinking about running races and obstacle races where users can view the video,
    take snapshots of the shots they like, and then purchase them. This would get around the 'missed shot' or
    the "he was blinking"

    So I started having the user interface developed. Allowing users to view videos and then capture images from them
    and hopefully purchasing them.

    Unfortunately I am still learning the video side. got to start somewhere

    The videos would be sourced/created by me or others working an event. Currently I have tried with the D800
    because that is what I own, but would be open after research to purchase other cameras that work better if need be.

    I have been in the event photo business for 15 years and have seen photo sales drop dramatically the last few years.
    My thought s that this is new , different and hopefully exciting to the users.

    I just need to get the video format/quality correct. And confirm the method it is being developed in will work (flash/flv)

    thanks

    John
    Quote Quote  
  11. Originally Posted by johnp12 View Post
    Hi,

    Ok, here is more background. I currently take and sell still photos at sporting events.
    We are often asked about offering video as well. My thought was to combine the 2. Offer both.
    Even though I know the frame grab quality will not be as good as a DSLR photo, but honestly,
    most people only are interested in 5x7 or 8x10's. So higher quality is overkill. and over time I am guessing
    video quality will only improve.

    I am also thinking about running races and obstacle races where users can view the video,
    take snapshots of the shots they like, and then purchase them. This would get around the 'missed shot' or
    the "he was blinking"

    So I started having the user interface developed. Allowing users to view videos and then capture images from them
    and hopefully purchasing them.

    Unfortunately I am still learning the video side. got to start somewhere

    The videos would be sourced/created by me or others working an event. Currently I have tried with the D800
    because that is what I own, but would be open after research to purchase other cameras that work better if need be.

    I have been in the event photo business for 15 years and have seen photo sales drop dramatically the last few years.
    My thought s that this is new , different and hopefully exciting to the users.

    I just need to get the video format/quality correct. And confirm the method it is being developed in will work (flash/flv)

    thanks

    John


    I like the idea, and hope you enjoy some success but there are some techical limitations you're going to experience

    I hope you understand that the print quality will be "terrible" compared to DSLR sourced still photos (even from jpeg, not raw) - even for 5x7 prints, even when taken from the original video. Video on the D800 is 4:2:0 chroma subsampled (1/4 of the color resolution), riddled with aliasing and moire beause the large sensor is "pixel binned" in video mode, also the compression is poor - full of compression artifacts. If you re-encode it to FLV in a bandwidth friendly encode to take screenshots, quality will drop even farther. (The D800 shoots ~24Mb/s, it's not "bandwith friendly" to stream, so even if you re-wrap it and lose no quality, you will have bandwidth concerns)

    The other problem you're going to have with the D800 is the videos are full range flagged - this means the black and white level will be crushed unless the screenshots , player and preview take this into account . If your programmer can develop this through quicktime instead of flash, it will help with the levels issue (quicktime decoder clamps the levels to "legal range")

    Quality wise, you're going to have the same concerns on most consumer and prosumer level cameras for video (most are subsampled, most will have compression issues) . Right now there is a huge gap between still photos (e.g. from DSLR) and video.

    I think you're going to have to make some tradeoff decisions regarding bandwith, ease of use, quality
    Quote Quote  
  12. thanks,

    I wonder if it makes sense then 2 have 2 files per video. One which the user views in flv format and when frames are grabbed, then take that frame from the 'hi res' video file (I hope I have the lingo correct). That would help the bandwidth issue.

    I'm not sure the cost of cameras that will do a better job (better than prosumer). Most DSLR camera we use now are around
    $3-5k. What recommendations would you have for cameras brands/models. (are 4k cameras better for this?)

    I really appreciate your help (with my learning process!)

    John
    Quote Quote  
  13. Originally Posted by johnp12 View Post
    thanks,

    I wonder if it makes sense then 2 have 2 files per video. One which the user views in flv format and when frames are grabbed, then take that frame from the 'hi res' video file (I hope I have the lingo correct). That would help the bandwidth issue.
    Actually I was going to suggest that; a lower bandwidth, lower quality "proxy" version for viewing in the application, and the original file to take the actual higher quality screenshots from .

    The reason I was reluctant to suggest this is frame accurate parsing will be a problem with your application. You might get the wrong frame. A user selects a frame , but it might not match up to the same frame on the "original" version. The problem is these compressed video codecs use long GOP encoding (temporal compression) as a method of reducing bandwith (each frame isn't complete, only differences between frames are stored) , and sometimes frame accuracy can be an issue when selecting a frame. The problem isn't there with intra codecs (all I frame)


    I'm not sure the cost of cameras that will do a better job (better than prosumer). Most DSLR camera we use now are around
    $3-5k. What recommendations would you have for cameras brands/models. (are 4k cameras better for this?)
    I'm not sure what "acceptable" quality means to you. There is a wide range of perception. It might be "good enough", but for a pro photographer, you should be able to see the difference in quality. Prints derived from typical HD video sources will look poor.

    Yes, the video quality will be substantially better with higher resolutions, even if they are chroma subsampled. The lowest priced decent model right now is Red Scarlet (body about $8K), and by the time you outfit all the accessories, lenses, recording media, etc... it's going to be easily >$20K . (There is a JVC model that does 4K, but it produces terrible quality, mushy video quality)

    Lower priced 4K video cameras are around the corner, the black magic 4k camera is about $4K body, and it can output prores (all I frame) and raw as options. On paper it looks fantastic , with a global shutter (no more rolling shutter artifacts!) - but nobody has actually used this or reviewed it, so I would still wait before placing a deposit
    Quote Quote  
  14. thanks


    so it sounds like if I get this developed using the 2 file method, over time the hi quality version of the video should just get better and better and more affordable and the software should not have to change.
    So using the 2 file method, would you recommend any certain video format as the 'raw' file?
    Quote Quote  
  15. Originally Posted by johnp12 View Post


    so it sounds like if I get this developed using the 2 file method, over time the hi quality version of the video should just get better and better and more affordable and the software should not have to change.
    It's too hard to say right now, because I suspect "flash" - the Adobe platform - will change soon, and there is increasing use of HTML5 . Also you might limit your potential audience with using flash (not officially compatible with some devices) - think of how people use smartphones, portable devices, etc... the proportion is only going to get larger

    There is new compression (HEVC) available which is better than the current h.264/AVC right now, and it's not clear how that fits into all this


    So using the 2 file method, would you recommend any certain video format as the 'raw' file?
    Always, the original file is the "best" . Some current problems are your "original" MOV from the D800 is already highly compressed, chroma subsampled (4:2:0) , and uses long GOP.

    For various camera models and recording devices - if you can get higher resolutions, less chroma subsampling (4:2:2 is better than 4:2:0), less compression, it will be better

    DSLR "video" typically have lower effective resolution than "traditional" video cameras. What I mean by this is most DLSR's produce "soft" filmic images in video mode. The actual measureable lines of resolution is only about 500-700 when measured on a test chart. So even though the frame size is 1920x1080, but the actual resolvable detail level is much less - you can think of it as the pixel quality being less sharp or not as good.

    4:2:0 Chroma subsampling means the color resolution is reduced in 1/2 in both dimensions. So for a 1920x1080 "video" , the color information is only 960x540 max.
    Quote Quote  
  16. thank you very much for your time. I really appreciate it
    Quote Quote  



Similar Threads

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