Realtime DV Processor at DVinfo.net

John Miller · June 7th, 2006, 06:41 AM

Enosoft's next product - the Enosoft DV Processor - will be available for public beta testing very soon!

What is the Enosoft DV Processor?

In short, a deceptively powerful tool for processing your DV footage. Key features include:

1. Realtime conversion of widescreen DV to letterboxed DV
2. Realtime 'burning' of time/date/timecode
3. Realtime logo insertion
4. Modify time/date/timecode encoded in the DV data
5. DV IN to DV OUT - no capturing necessary. If you have two DV devices, simply put your source material in one and, in realtime, output the processed DV to the other device.
6. Capture from DV to AVI (Type 1 or Type 2)
7. Realtime processing of AVI to DV OUT
8. AVI to AVI processing
9. Scene detection

i.e., the processor's input can be an AVI file or live DV device; the output can be an AVI file or live DV device.

Imagine! You have widescreen footage and need a letterboxed copy. The usual way to do this is capture the footage, convert in an NLE, send converted footage to device. For a 30 minute program, this might take 30 minutes to capture, an hour to convert and another 30 minutes to send back to the device. With the Enosoft DV Processor, just press play on one device and see the converted footage on the other device - live!

SYSTEM REQUIREMENTS: Windows XP. Some features require SSE2-capable 32-bit processor. Without SSE2, these will not be available.

Reference System: Thinkpad T40 laptop - Pentium M 1.5GHz CPU, 1GBMB RAM, 40GB 5400rpm HDD, 32MB ATI Radeon 7500, WinXP Pro SP2, OHCI IEEE-1394 Cardbus

To take part, please sign up for Beta Testers membership at the Enosoft Users Group. (You will need to register for the main group at http://www.enosoft.net/phpBB2/profile.php?mode=register, log-in, click on Usergroups (http://www.enosoft.net/phpBB2/groupcp.php) and request memberships for the Beta Testers group).

Peter Jefferson · June 7th, 2006, 07:42 AM

interesting stuff... just afew notes/questions...

1. Realtime conversion of widescreen DV to letterboxed DV

((Some might need this, however this seems to be a cropping as opposed to an aspect ration change?? Pls correct me if im wrong. As an aside, wouldnt it be more of a benefit to offer 4:3 conversion to 16:9 on the fly?? that would sell more units IMO

2. Realtime 'burning' of time/date/timecode

((Cool))

3. Realtime logo insertion

((can this be manipulated to act as a transparent watermark on footage? Lets say a company logo in white, running at 50% opacity, this way the footage isnt affected all that much... ))

4. Modify time/date/timecode encoded in the DV data

((Cool.. this would save all those unsyncronised cameras running different timestamps.. lol for those doing multicam work with different companies... like one time i did a stage show with 12 cameras... 4 were our own, the other 8 were from different companies and shooters. Its amazing how many ppl DONT set up their cameras properly))

5. DV IN to DV OUT - no capturing necessary. If you have two DV devices, simply put your source material in one and, in realtime, output the processed DV to the other device.

((This is already possible through windows without any SW. Basically what your saying is that i can use this to dupe one tape from one device to another?? This is already possible either through using windows as an interface, or via direct camera 1394 connection. If im wrong please correct me, im jsut tryin to understand what it does. ))

6. Capture from DV to AVI (Type 1 or Type 2)

((yup))

7. Realtime processing of AVI to DV OUT

((Groovy))

8. AVI to AVI processing

((??? ))

9. Scene detection
((cool.. ))

One question though, does this alow for multiple instance of the program to be run and to allow for multiple streams of DV to be captured via 2 or 3 different devices?? Will it recognise that there are 3 cameras of the SAME MAKE and segregate each stream?? Scenealyzer claims to be able to do this, and some people have had success, however i have never been able to get this going without crashing the system.
If ur app can capture multiple streams, then ur in with a winner

John Miller · June 7th, 2006, 09:08 AM

Hi Peter,

Thanks for your comments. Please see my responses below.

John.

Quote:

Originally Posted by Peter Jefferson

interesting stuff... just afew notes/questions...

1. Realtime conversion of widescreen DV to letterboxed DV

((Some might need this, however this seems to be a cropping as opposed to an aspect ration change?? Pls correct me if im wrong. As an aside, wouldnt it be more of a benefit to offer 4:3 conversion to 16:9 on the fly?? that would sell more units IMO

It isn't cropping. It's proper squeezing of a native 16:9 image so that, when viewed on a standard 4:3 TV, it will look correct. In NLE terms, it would be like using Premiere to rescale vertically by 75%.

Are you refering to faked widescreen (the way most camcorders do it), whereby the image is simply cropped to give the appearance of widescreen? If so, do you think there is a lot of need to remove the cropping and rescale? That's possible (but not a current feature).

2. Realtime 'burning' of time/date/timecode

((Cool))

3. Realtime logo insertion

((can this be manipulated to act as a transparent watermark on footage? Lets say a company logo in white, running at 50% opacity, this way the footage isnt affected all that much... ))

Absolutely! - you'd provide the logo as a bitmap with an alpha channel to determine the opacity OR specify the opacity of the whole image in the program

4. Modify time/date/timecode encoded in the DV data

((Cool.. this would save all those unsyncronised cameras running different timestamps.. lol for those doing multicam work with different companies... like one time i did a stage show with 12 cameras... 4 were our own, the other 8 were from different companies and shooters. Its amazing how many ppl DONT set up their cameras properly))

Also useful for capturing live feeds where the date/time/timecode aren't present. You can add the current time/date, for example.

5. DV IN to DV OUT - no capturing necessary. If you have two DV devices, simply put your source material in one and, in realtime, output the processed DV to the other device.

((This is already possible through windows without any SW. Basically what your saying is that i can use this to dupe one tape from one device to another?? This is already possible either through using windows as an interface, or via direct camera 1394 connection. If im wrong please correct me, im jsut tryin to understand what it does. ))

Not just straight duplication but doing the aspect ratio conversion, time/date burning, logo insertion etc. i.e., the input signal is processed and fed to the output device. No need to capture the source, render it in an NLE and send back to the device.

6. Capture from DV to AVI (Type 1 or Type 2)

((yup))

7. Realtime processing of AVI to DV OUT

((Groovy))

8. AVI to AVI processing

((??? ))

If the processing is too demanding and can't be achieved in realtime, you can do it from AVI to AVI (like a regular editing app). Also allows Type 1 to Type 2 conversion (and vice versa).

9. Scene detection
((cool.. ))
And with the option to burn the tape name/ID, scene name/ID, position within scene onto the video

One question though, does this alow for multiple instance of the program to be run and to allow for multiple streams of DV to be captured via 2 or 3 different devices?? Will it recognise that there are 3 cameras of the SAME MAKE and segregate each stream?? Scenealyzer claims to be able to do this, and some people have had success, however i have never been able to get this going without crashing the system.
If ur app can capture multiple streams, then ur in with a winner

Right now, it won't do it but we have experimented with this. A single instance of the program could have multiple input streams and the captured AVI will contain the same number of DV streams. Obvious benefits here are to ensure synchronization etc.

Note also that the input and output devices are controllable from on-screen, including shuttling at different speeds etc.

I'll put a couple of images up on our site and post a link to them so you can see what the user interface is like (at the moment!) and an example of a before and after processed video frame

John Miller · June 8th, 2006, 03:17 PM

Quote:

Originally Posted by Peter Jefferson

interesting stuff... just afew notes/questions...

One question though, does this alow for multiple instance of the program to be run and to allow for multiple streams of DV to be captured via 2 or 3 different devices?? Will it recognise that there are 3 cameras of the SAME MAKE and segregate each stream?? Scenealyzer claims to be able to do this, and some people have had success, however i have never been able to get this going without crashing the system.
If ur app can capture multiple streams, then ur in with a winner

UPDATE! I tried this today and it worked like a charm. The set up was:

1 x Sony DSR-PDX10 in playback mode
1 x Sony DCR-TRV50 in camera mode
1 x External USB 2.0 drive for capturing
Thinkpad T40 laptop (1.5GHz Pentium M single core, 1GB RAM)
2 x Instances of the Enosoft DV Processor both set up to burn the time/date/timecode into the captured DV files

System throughput was excellent - no dropped frames inspite of writing two large files to the same drive at the same time!

I then opened both files in GraphEdit (a Microsoft freebie tool for developers) and happily played both files back.

During the capture, I used the camcorder in camera mode to record the LCD display of the one playing a tape.

I have put two screen shoots on our site:

1. The two instances of the processor, configured for the two cameras, ready to capture/convert (the capture hadn't started, so the "Output" side of the processor is empty). Note the device control buttons for the device in playback mode - full control from the computer.

http://www.enosoft.net/DVProcessor/T...20To%20Run.gif

(Image size: 61kB - a full size 1024 x 768 screenshot)

2. The GraphEdit example playing both captured files. Note both playback windows show the burned-in time/date/timecode. For the camera mode device (i.e., a live stream) there are no data...

http://www.enosoft.net/DVProcessor/T...%20Playing.gif

(Image size: 146kB - a full size 1024 x 768 screenshot)

If you look closely enough, you'll see that the image of the LCD display is not exactly the same as the captured video from that camcorder. This is because there was about a two second delay between starting the two processors. It would be a trivial matter to have one instance synchronize the other so both start together.

I think I'll give three devices a try...(I have another Sony somewhere).

If you want to try it out, please sign up for the beta testing!

Peter Jefferson · June 10th, 2006, 09:49 AM

"It isn't cropping. It's proper squeezing of a native 16:9 image so that, when viewed on a standard 4:3 TV, it will look correct. In NLE terms, it would be like using Premiere to rescale vertically by 75%."

((OK cool, but how about the horizontal aspect? We may easily stretch vertically, however teh aspect would still be incorrect unless its tagged. THEN to fill out the vertical bars, we'd need to scale the 4:3 stretched frame to FILL the frame (i'll email you when i get a chance to do some screen dumps to show u what im rambling on about)
So not only must we squeeze vertically, but we must also fill out the the frame, else we'd be left with a 16:9 tagged frame, 16:9 ACTUAL frame, with 4:3 centred media, with black bars on the sides AND the top and bottom.. much like 2.35.1 aspect but without the width.. am i making any sense?? Like i said i'll email u some dumps of what im talkin about.. ))

"Are you refering to faked widescreen (the way most camcorders do it), whereby the image is simply cropped to give the appearance of widescreen? "

((Nah, this can be done easily with a PNG mask overlay or pan crop of the clip.. ))

"If so, do you think there is a lot of need to remove the cropping and rescale? That's possible (but not a current feature)."

((I think there IS a need for 4:3 to 16:9 squeezing.. definately.. however at this time, not many programs do this in realtime.. especially during capture.
The probelm with this though is that most shots are framed for 4:3, so by doing it in realtime, without review, u might inadvertantly crop and stretch a shot which needs reframing (lifting or lowering of teh WHOLE FRAME, PRIOR to 16:9'ing (there IS a 1 in front of that.. lol)

Does that make sense??
Lets say i have a talking head shot.. and in 4:3 ive framed his head jsut above the eyes, centred between his eyes and his forhead/hairline.
Now for 4:3, this is fine... BUT if i was to crop and squeeze, about 12% of my headspace and neckspace will be discarded... now when looking at THIS after cropping and squeezing, the 16:9 frame forces his eyes to be too cloose to the top of my frame..

Does that make a little more sense now?? This is why i think aspect conversion MAY work for most shots, but to have the option of tweakin it for each respective clip (whether it during capture, or transfer back to tape/file) would be beneficial.. Having tha t option makes aspect conversion and render headaches obsolete.. and if the footage is SHOT as letterbox, this framing issues wont be an issue.. but for full frame 4:3 to 16:9, then it COULD be..

John Miller · June 11th, 2006, 07:04 AM

I've put a couple of still frames on our site showing pre- and post-processing for aspect ratio conversion.

Pre: http://www.enosoft.net/DVProcessor/native16_9.JPG
Post: http://www.enosoft.net/DVProcessor/letterboxed.JPG

Note, both images are snapshots from Windows Media Player using our DV decoder which overlays the text information etc. As you can see, all of the embedded information is maintained during the conversion. In the bottom right corner you can see the aspect ratio is retagged appropriately.

Another couple of images are also online that show that interlaced footage is processed correctly (simply rescaling a 720x480 or 720x576 image to 720x360 or 720x432 will NOT preserve interlacing).

Pre: http://www.enosoft.net/DVProcessor/native16_9_2.JPG
Post: http://www.enosoft.net/DVProcessor/letterboxed_2.JPG

The original footage was recorded using a Sony DSR-PDX10 in true 16:9 mode. The converted footage will appear "correct" on a standard 4:3 TV.

The original incentive to develop the realtime aspect ratio conversion was that I have accumulated many hours of widescreen DV footage and, much to my dismay, widescreen TVs in North America are all but non-existent OR very expensive (generally, you have to buy an HD set to get the widescreen). This is different than, say, Europe, where widescreen standard definition sets have been available for years. The realtime conversion means that I can record something widescreen and, immediately, view the results on a standard TV in the right way...rather than go through hours and hours of capture/render etc.

June 7th, 2006, 06:41 AM	#1
John Miller Major Player Join Date: Oct 2005 Location: Hillsborough, NC, USA Posts: 968	Realtime DV Processor Enosoft's next product - the Enosoft DV Processor - will be available for public beta testing very soon! What is the Enosoft DV Processor? In short, a deceptively powerful tool for processing your DV footage. Key features include: 1. Realtime conversion of widescreen DV to letterboxed DV 2. Realtime 'burning' of time/date/timecode 3. Realtime logo insertion 4. Modify time/date/timecode encoded in the DV data 5. DV IN to DV OUT - no capturing necessary. If you have two DV devices, simply put your source material in one and, in realtime, output the processed DV to the other device. 6. Capture from DV to AVI (Type 1 or Type 2) 7. Realtime processing of AVI to DV OUT 8. AVI to AVI processing 9. Scene detection i.e., the processor's input can be an AVI file or live DV device; the output can be an AVI file or live DV device. Imagine! You have widescreen footage and need a letterboxed copy. The usual way to do this is capture the footage, convert in an NLE, send converted footage to device. For a 30 minute program, this might take 30 minutes to capture, an hour to convert and another 30 minutes to send back to the device. With the Enosoft DV Processor, just press play on one device and see the converted footage on the other device - live! SYSTEM REQUIREMENTS: Windows XP. Some features require SSE2-capable 32-bit processor. Without SSE2, these will not be available. Reference System: Thinkpad T40 laptop - Pentium M 1.5GHz CPU, 1GBMB RAM, 40GB 5400rpm HDD, 32MB ATI Radeon 7500, WinXP Pro SP2, OHCI IEEE-1394 Cardbus To take part, please sign up for Beta Testers membership at the Enosoft Users Group. (You will need to register for the main group at http://www.enosoft.net/phpBB2/profile.php?mode=register, log-in, click on Usergroups (http://www.enosoft.net/phpBB2/groupcp.php) and request memberships for the Beta Testers group).

June 7th, 2006, 07:42 AM	#2
Peter Jefferson Inner Circle Join Date: Apr 2003 Location: Aus Posts: 3,884	interesting stuff... just afew notes/questions... 1. Realtime conversion of widescreen DV to letterboxed DV ((Some might need this, however this seems to be a cropping as opposed to an aspect ration change?? Pls correct me if im wrong. As an aside, wouldnt it be more of a benefit to offer 4:3 conversion to 16:9 on the fly?? that would sell more units IMO 2. Realtime 'burning' of time/date/timecode ((Cool)) 3. Realtime logo insertion ((can this be manipulated to act as a transparent watermark on footage? Lets say a company logo in white, running at 50% opacity, this way the footage isnt affected all that much... )) 4. Modify time/date/timecode encoded in the DV data ((Cool.. this would save all those unsyncronised cameras running different timestamps.. lol for those doing multicam work with different companies... like one time i did a stage show with 12 cameras... 4 were our own, the other 8 were from different companies and shooters. Its amazing how many ppl DONT set up their cameras properly)) 5. DV IN to DV OUT - no capturing necessary. If you have two DV devices, simply put your source material in one and, in realtime, output the processed DV to the other device. ((This is already possible through windows without any SW. Basically what your saying is that i can use this to dupe one tape from one device to another?? This is already possible either through using windows as an interface, or via direct camera 1394 connection. If im wrong please correct me, im jsut tryin to understand what it does. )) 6. Capture from DV to AVI (Type 1 or Type 2) ((yup)) 7. Realtime processing of AVI to DV OUT ((Groovy)) 8. AVI to AVI processing ((??? )) 9. Scene detection ((cool.. )) One question though, does this alow for multiple instance of the program to be run and to allow for multiple streams of DV to be captured via 2 or 3 different devices?? Will it recognise that there are 3 cameras of the SAME MAKE and segregate each stream?? Scenealyzer claims to be able to do this, and some people have had success, however i have never been able to get this going without crashing the system. If ur app can capture multiple streams, then ur in with a winner

June 10th, 2006, 09:49 AM	#5
Peter Jefferson Inner Circle Join Date: Apr 2003 Location: Aus Posts: 3,884	"It isn't cropping. It's proper squeezing of a native 16:9 image so that, when viewed on a standard 4:3 TV, it will look correct. In NLE terms, it would be like using Premiere to rescale vertically by 75%." ((OK cool, but how about the horizontal aspect? We may easily stretch vertically, however teh aspect would still be incorrect unless its tagged. THEN to fill out the vertical bars, we'd need to scale the 4:3 stretched frame to FILL the frame (i'll email you when i get a chance to do some screen dumps to show u what im rambling on about) So not only must we squeeze vertically, but we must also fill out the the frame, else we'd be left with a 16:9 tagged frame, 16:9 ACTUAL frame, with 4:3 centred media, with black bars on the sides AND the top and bottom.. much like 2.35.1 aspect but without the width.. am i making any sense?? Like i said i'll email u some dumps of what im talkin about.. )) "Are you refering to faked widescreen (the way most camcorders do it), whereby the image is simply cropped to give the appearance of widescreen? " ((Nah, this can be done easily with a PNG mask overlay or pan crop of the clip.. )) "If so, do you think there is a lot of need to remove the cropping and rescale? That's possible (but not a current feature)." ((I think there IS a need for 4:3 to 16:9 squeezing.. definately.. however at this time, not many programs do this in realtime.. especially during capture. The probelm with this though is that most shots are framed for 4:3, so by doing it in realtime, without review, u might inadvertantly crop and stretch a shot which needs reframing (lifting or lowering of teh WHOLE FRAME, PRIOR to 16:9'ing (there IS a 1 in front of that.. lol) Does that make sense?? Lets say i have a talking head shot.. and in 4:3 ive framed his head jsut above the eyes, centred between his eyes and his forhead/hairline. Now for 4:3, this is fine... BUT if i was to crop and squeeze, about 12% of my headspace and neckspace will be discarded... now when looking at THIS after cropping and squeezing, the 16:9 frame forces his eyes to be too cloose to the top of my frame.. Does that make a little more sense now?? This is why i think aspect conversion MAY work for most shots, but to have the option of tweakin it for each respective clip (whether it during capture, or transfer back to tape/file) would be beneficial.. Having tha t option makes aspect conversion and render headaches obsolete.. and if the footage is SHOT as letterbox, this framing issues wont be an issue.. but for full frame 4:3 to 16:9, then it COULD be..

June 11th, 2006, 07:04 AM	#6
John Miller Major Player Join Date: Oct 2005 Location: Hillsborough, NC, USA Posts: 968	I've put a couple of still frames on our site showing pre- and post-processing for aspect ratio conversion. Pre: http://www.enosoft.net/DVProcessor/native16_9.JPG Post: http://www.enosoft.net/DVProcessor/letterboxed.JPG Note, both images are snapshots from Windows Media Player using our DV decoder which overlays the text information etc. As you can see, all of the embedded information is maintained during the conversion. In the bottom right corner you can see the aspect ratio is retagged appropriately. Another couple of images are also online that show that interlaced footage is processed correctly (simply rescaling a 720x480 or 720x576 image to 720x360 or 720x432 will NOT preserve interlacing). Pre: http://www.enosoft.net/DVProcessor/native16_9_2.JPG Post: http://www.enosoft.net/DVProcessor/letterboxed_2.JPG The original footage was recorded using a Sony DSR-PDX10 in true 16:9 mode. The converted footage will appear "correct" on a standard 4:3 TV. The original incentive to develop the realtime aspect ratio conversion was that I have accumulated many hours of widescreen DV footage and, much to my dismay, widescreen TVs in North America are all but non-existent OR very expensive (generally, you have to buy an HD set to get the widescreen). This is different than, say, Europe, where widescreen standard definition sets have been available for years. The realtime conversion means that I can record something widescreen and, immediately, view the results on a standard TV in the right way...rather than go through hours and hours of capture/render etc.