So you think you'd like to contribute some raw materials for the artists
to work on? Here's advice.
Scanning
Capping
Capping
Hardware
So you think you'd like to contribute some raw materials for the artists
to work on? Here's advice.
Scanning
Capping
Capping
Hardware
Remember the watchword: click on the thumbs to see a bigger version - or do "open in new window" to get it side by side with the text.
Subject: Re: lara 4 and 4a
Date: Tue, 3 Jul 2001 22:25:04 +0100
From: "Michael Quick" <mike@mquick.freeserve.co.uk
Newsgroups: alt.binaries.multimedia.xena-herc
| As to tips, I've been following this procedure
with my scans :
1. scan at 300dpi, and try placing a black sheet of paper behind the pic to stop the scanner 'seeing' the other side of the paper (if scanning magazines) 2. trim border to eliminate any areas outside the main picture 3. resize the longer side to 1200 pixels (I know, my scans are too big!) 4. in Photoshop, zoom in 500% and slowly go over the entire scan cloning out all defects (and there will be some!). |
5. use PS's Levels command to play with the Contrast/Brightness.
All scans look better with a slight tweak using this particular command.
If your scanner is set up correctly, you might not need to do this.
Still worth trying though...
6. apply a very mild Unsharp filter (perhaps 50% amount, 2 pixels radius and 3 thresholds) And that's that :) |
Let's start with some general advice from the always helpful Mike Quick:
(Click on the Thumbs to see larger examples)
| PC set-up:
First you should ensure the monitor is correctly set up. 1. set the background to plain black 2. set the contrast setting to maximum 3. set the brightness to dark, now slowly turn up the brightness until the background turns slightly grey. Now set the brightness back one notch to regain a truly black background. That should ensure that the monitor is displaying what most other people are seeing. Otherwise you might be 'correcting' an image that's just fine, to one that looks good only on your PC! ------------------------ Faults There are a multitude of possible faults with screencaps! Amongst them are : 1. Interlace 2. Interference lines 3. Colour saturation 4. Poor contrast 5. Dot crawl 6. Motion break-up 7. Over compressed caps |
Some solutions
<g>
Interlace: TV pictures are transmitted either 30 times (NTSC) or 25 (PAL) per second, BUT each frame is sent in two passes ('fields'). The first time all the even numbered lines are shown, then on the next pass all the odd numbered lines are drawn. Thus the term 'interlace'. The two 'fields' are interlaced together to give the impression of a single image. It's done this way to reduce the cost of the TV which can't otherwise keep up with the rate the images are broadcast (slight simplification here). Computer monitors are built to higher standards and can display non-interlaced pictures at up to 100 images per second (the 'refresh' rate you might have seen mentioned when talking about graphics cards and monitors) Symptoms : Horizontal black lines or two different images intermixed. Solution : Get a deinterlace filter! A lot of programs will accept the standard Photoshop (PS) plugin filters. Paint Shop Pro (PSP) will for example as does Corel Paint (CP). If the interlace is only partial or slight you can sometimes get rid of it by drawing a selection around the area with the 'Lasso' tool and then apply a 'Motion Blur' to it. Set it to vertical with a blur of one pixel. That might help. I use the deinterlace filter mostly on 'action' caps. When
the images on the TV picture alter rapidly you're most likely to get interlace.
After using a de-interlace filter you usually have to go over the de-interlaced
areas with the 'Smooth' tool to reduce the 'orange peel' effect it leaves
behind.
|
||
|
|
|||
| Interference
lines
Symptoms : These are horizontal lines caused by signal interference. If you check my Fallen Angel caps on the site (http://www.mikes-images.com/) you'll see the left over remnants of lots of them! Solution : The easiest method to solve the problem is to avoid it in the first place! I got rid of them by replacing the cable that takes the signal to my card with a better shielded cable (called 'composite' in Europe and 'RF' in the States). They are available in various stores (including audio shops). The other means of removing them is by using the clone tool. Set it to a small size and about 50 to 75% strength. Slowly copy good areas of the cap over the top of the bad areas. Then use a very weak (25%) 'Smooth' tool afterwards over the affected areas. I did a lot of this on the Fallen Angel caps, they took about
15 to 20 minutes to finish each.
|
Colour
saturation
Symptoms : When flesh looks orange or red (high saturation). When flesh looks white or very pale (low saturation). Solution : Reduce/increase the saturation. There should be a command called 'Saturation' somewhere. In PSP it's under 'Colors>Adjust>Hue/Saturation/Luminence'. Just put in a lower setting for saturation until the cap looks right. And, of course, if the saturation is too low, you can just reverse the
above.
|
||
|
|
|||
| Poor contrast
Symptoms : caps look pale, washed out, grey... (low contrast - most common) or very dark (high contrast - uncommon) Solution : Reduce/increase the contrast. There should be an obvious Contrast command in the menus. Alter the setting until shadow areas turn black (if grey previously). Sometimes adding brightness and contrast simultaneously to the cap can improve it. Adding slightly more brightness than contrast works well, such as +5% Brightness and +3% Contrast Dark caps should be jet black in the shadow areas, not grey.
|
Dot crawl
Symptoms : particularly in red areas the pixels 'clump' together in small darker dots. Seems to be a result of the compression software used to broadcast digital TV. Solution : I use the smooth tool
set to a small size, weak strength (30%) and work over the affected areas
smoothing the dots out.
|
||
|
|
|||
| Motion
break-up
Symptoms : total garbage on the screen! Occurs mostly when capturing fast action scenes. This is why you see so few action caps. Some capture cards can be set to capture 'one field' only. This can reduce the problem. Solution : Throw the cap away :(
|
Over compressed
caps
Symptoms : compression artefacts (pixels tend to clump together in 8 x 8 groups) show up when zooming into the image. Solution : If they are as a result of your own caps then change the setting that determines the amount of compression. A 640 x 480 cap (typical for NTSC) should be about 50-60k after compressing (down from 921k uncompressed) If they're in other peoples caps then the only course of action is to
either blur the images to smooth the artefacts together (and lose sharpness/detail),
or to go over the cap with the smooth tool (set to a low strength - 25%
or so) getting rid of the worst areas only.
|
||
|
|
|||
And, here's a discussion of Hardware, and using same.
Warning: Very technical stuff ahead. Not for the faint of heart. [smile]
Subject: Re: opinions on video capturing?
Date: Sun, 28 Jan 2001 02:32:19 GMT
From: Dmac <dmac2@mediaone.net>
Newsgroups: alt.binaries.multimedia.xena-herc
| Wow. A lot of info. I saw your card at my local
Staples for $49. I think I'll give it a try. I'm getting digital cable
installed next week so I think I will be able to post my own caps without
being embarassed of them. Hell, if your card doesn't do the job I will
have only lost $49. Thanks for the info.
Dmac Hi, a few comments... Hardware. capture cards come in three types: 1. A dedicated capture card such as my Hauppauge WinTV card or Hubbes Pinnacle DC-30 card (I think I remember correctly). The WinTV card has the benefit of being just about the cheapest capture card that you can find (about £40/US$60). The DC-30 is about £600/US$900, but very high quality. 2. A graphics card that also has a TV tuner on the card. Typical examples are the ATI All-In-Wonder and the Matrox G400/G450 Marvel. The newer Radeon AIW card is pretty expensive though, about £280/US$420. I also understand there are considerable problems with drivers at the moment. I know that both Psyche and Amazon have experienced troubles. Artemis/Xenaholic have just bought a Dazzle capture card, but are experiencing a few teething problems at the moment. I have no knowledge of this card. 3. A plug in capture card. These can plug into either the parallel port or a USB port. Starman uses a device called Snappy that uses the parallel port. My WinTV card can be bought in a version that plugs into a USB port. Costs slightly more though (£70/US£105) ... Signal. It's a sad fact that about 70% of the final quality is entirely down to a good signal, the remainder is dependant on your skill with a graphics program to clear up faults on the caps. ... Here's a few further comments about the only card I know well, my one, the WinTV card. |
Capture size varies from 640 x 480 up
to a massive 1600 x 1200 - although as it's done in software you may as
well capture at 800 x 600 and resize it yourself in PhotoShop.
You can choose to capture only one field - useful for capturing moving images (all TV pictures are transmitted twice, once showing all the even numbered lines then again for the odd numbered lines. Thus the term 'Interlaced', the two fields are interlaced together to form a single picture. However, sometimes the two fields don't match during very rapid movement and thus you get interlacing showing on the image. A bad fault. You can capture video, but only at a low resolution, I think 352 x 288 max. You can capture live from broadcast or from tape. Live broadcast gives a massively better capture. You can choose the type of input signal - RF (TV aerial cable), Composite (I think this is called RCA in the North America) which I use and S-Video (the best I think). You can get the TV picture to fill the entire screen, in which case it looks just like a normal TV screen. Some models have a remote control, so it would then function identically to a normal TV. ... Attached is a screenshot of my WinTV capturing some pics from the Olympics. [sorry, missing - ed.] Note that the picture is being captured at 800 x 600, both fields.
Saving to file can be a little slow. On average, over an entire
program you can capture at an average of one pic
The AIW cards and the Matrox cards work differently. They capture the entire episode or TV show to disk as a Video file, then you extract the images you want at a later date. This has a big advantage in that ALL the possible images are available to choose from (30 NTSC images per second x say 45 minutes x 60 seconds = 81,000 images!) One problem with this approach is that you need a very large, very fast hard disk! Storing 80,000+ images takes up a huge amount of space. Spencer uses the Matrox card and has twin fast 30gb drives which are kept virtually empty prior to capturing. |
Subject: Re: First time screencapping..
Date: Wed, 21 Feb 2001 06:40:30 -0700
From: Starman <starman@!buzzoff!connect.ab.ca
Newsgroups: alt.binaries.multimedia.xena-herc
| On Wed, 21 Feb 2001 07:13:47 GMT, Jen <AlaniScully@strangejourney.com
wrote: Well I finally got around to using my Snappy. I had taped Path
Of Vengeance but the quality is kinda bad. I had The Ring on another tape,
better quality there so I took a few caps. I took a few of the credits
too. I thought a few of these came out good anyway. Tell me if these are
any good and if I should keep on posting. And suggestions are definitely
welcome!
As a user of Snappy, this is one of the few times I may be in the best position to help. My PoV Livia pic, just a few messages away, was captured with Snappy (off a live digital satellite signal). I am commenting here rather than by mail for the possible benefit of others who may be interested in making vidcaps, but did not take the plunge like yourself. 1. You will notice there are black horizontal lines on all the pics. If this is what I think it may be, you can't get rid of it. For some reason, certain channels produce this effect (could be a Snappy or vcr quirk). I have never found a cure and it is extremely difficult to edit out. Best bet if it persists is to find another channel that XWP is on. If no alternatives, you may be sol. You will find that you get far better pics capturing live rather than from tape. Snappy was designed more for work with tape or home video camera (very slow operation) but tape certainly does not give optimum results. The sharpness from typical vcrs is quite pathetic and tape introduces a bunch of other problems too - color banding, graininess, flat colours. Tape does have the advantage you don't have to be there during the show! 2. Your exposure seems way off. You should not have large areas of pure white. Often it is bound to happen in highlights eg sunshine on hair but one should aim for the very brightest area being just white and the darkest shadow being black. "Credits Xena1" is probably the best pic. Mike Quick posted some info not so long ago on checking pics using the histogram available on many graphics editors. Attached is the histogram for one of your pics. It shows the distribution from black (left) to white (right). You can see there is very little black and loads of white. The histogram does not have to be flat but should be more even and spread out. Pic #2 has been corrected by me but the excess white cannot be removed completely. Even if you darken, you just have excess gray instead - see next para. Once you have overexposed the bright areas (saturated to white 255), you cannot reverse the effect by editing. If you underexpose, the dark areas are saturated to black 0 and this cannot be reversed. It is therefore important to get things fairly correct at capture. The eye is not a bad judge, providing your monitor is set up ok. If your monitor is set very dark, you will have to increase brightness so it looks ok. When seen on a brighter monitor, it will look too bright. My personal recommendation here is to compare with typical pics in the ng. Your pics should look similar. Beware, Legman is brighter than myself, Mike, Psyche and Spencer, Dmac is very dim (referring to pics of course!). With Snappy, check whether you have "connected to tv" selected in the setup. If you have, and it is not connected to a tv, the caps will be too bright. I always use a small tv connected to the video out and capture in the "adjust" mode (I have v 2.0) because it is easier to quickly adjust controls. Unfortunately the spacebar "snap" does not work in this mode, you have to use a mouse. |
Reduce brightness and adjust contrast so not
too much black or white. Also, "picture", which is the gamma setting,
will greatly affect the pic. This determines where the colors will
lie relative to black and white. High "picture" = pale bright image.
I mostly set to +11 but that may depend on tv/vcr signal etc so experiment.
3. Your skin tones are mostly yellow - not too bad. That is probably the way the signal arrives. Yellow is fine in artificial light (such as candlelight in many xwp scenes). It should not be prominent in daylight - skin can still be yellowish though. If I have a slightly yellow signal, I set "tint" to -1 or -2. If the signal is extremely yellow I may set in addition, "red" +1 and "green" -1. Check other colors to ensure they are not made incorrect. Sometimes you have to compromise. 4. You are capturing at 640x480. In theory all the info for NTSC transmissions should be caught. I prefer to capture larger - more on that later. You always get some edge distortions which ideally are removed (usually worst top or bottom). Not everyone bothers, particularly on small pics. You should find the garbage is always in the same place (for example 3 pixels at the top, 15 at the bottom, 5 either side). There are several ways to eliminate but all require some editing knowledge. First is to select the "good" area of pic and resize to completely fill the frame. Second, if you only have a few pixels of distortion copy the adjacent rows or columns over the top. Third, capture oversize and crop the edges to get 640x480. I use fairly advanced CorelScripts to crop my pics automatically after capture so a 1048x786 pic is cropped to 1024x768, edge distortions gone. I know from experience where the garbage will be. 5, Capturing at larger size is slower and in theory a waste of time. However, Snappy smarts seem to be able to get the most out of the information and the more pixels you have, the easier it is to edit. You can reduce size later. You might want to see how much time it takes to capture and save different sizes. Snappy is quite slow so you may still prefer to capture small for speed (if working live) or because you only want small pics. Probably not worth capturing more than one "size" above final pic anyway. Eg 800x600 to produce 640x480. 6. Whether you edit off the tv station logo is personal preference. Not usually worth the bother on a small pic especially if you are only going to use parts of pics for montages etc. If you do want them off, use a clone tool. Requires lots of practice but not that difficult. 7. The detail in the pics is not bad and would be even better with correct contrast and brightness. With a bit of adjustment great improvements could be made. The horizontal lines may be the most troublesome fault. 8 You have probably already found the quickest way to capture is to save as soon as the snap is taken (don't wait for Snappy to create the full size image). Then edit later in a grapihics editor. If you are using Snappy to correct the pic after capture, then you have to wait and use the full image. I think that lot will do for now! Hope it's of some use. A good first effort. Keep at it because that's the only way you get experience and better. Starman.
|
Before:
 |
After:
 |
Subject: Re: screenshots ?
Date: Sun, 04 Mar 2001 12:37:07 -0700
From: Starman <starman@!buzzoff!connect.ab.ca>
Newsgroups: alt.binaries.multimedia.xena-herc
| On Sun, 4 Mar 2001 12:43:42 +0100, "Akazukin
Chacha"
<Charlotte.Welle@village.uunet.be> wrote: Hello I can see you take great pictures from the show. I believe
you take it with your pc but I'm still wondering how you do to have so
great quality pictures ! In fact, I try (with a miro pctv and an
Ati all-in-wonder) to do some screenshot but the quality is lower and in
action (xena battle for instance) is blurred and unusable... So someone
could help me ?
Quality in General. The most important factor is the tv signal that you get. If it's poor you are unlikely to get good vidcaps. You can do quite a lot after capture by editing in a graphics editor but that greatly reduces the number of good pics you can produce. If you captures are as good as the tv picture, that's all you are going to get. Usually the captures look worse than the live picture (persistence of vision) but occasionally they come out better. I vidcap Star Trek TOS which is average quality live but the vidcaps are excellent. Most serious vidcappers are now using digital tv. Legman in this ng is still on analog but has an exceptionally good signal. Digital is still no guarantee of good pictures as it has it's own set of problems. If you have intervening hardware such as cable or satellite receivers, vcrs etc, make the connections with composite or s-video (best) and keep the cable runs short. I think people get too worked up about weak signals. These days cable in urban areas and satellite are normally good. If your tv pic looks ok, your signal is ok. Video capture devices are not deliberately made to function worse than your tv! If the tv pic is bad, get your cable company in or try one of those really cheap signal amps. It helps to have the largest selection of channels that carry your favourite shows. You may find some give a better picture than others and the size, duration and ugliness of the station logo varies considerably. Some channels display their logo all the time (except the ads), some just after each break, some rarely. The less obtrusive the logo the less editing required if you want the logos off. I think you will find that almost any vidcap to be posted at 800x600 or greater needs some manual editing to remove defects. What you see posted in this ng is not exactly what was captured, in fact may be quite different. I posted an action picture yesterday of an Amazon flying through the air. Any similarity with what was captured is purely coincidental (!) since the vidcap was totally pixelated. Spending some time with a graphics editor will improve things. I have not seen too many good large pics captured with an Ati all-in-wonder. I don't know much about the Miro pctv. Mike uses a Hauppage Win Tv card which is dirt-cheap but seems to give excellent results (partly because the software screwed itself up!) so money is not everything. Action Pics Capturing pictures on tv is no different to action photography. For action photography, you would not dream of shooting at less than 1/250th sec and |
typically one uses 1/500th or 1/1000th sec.
A video capture is nominally 1/25th sec (Europe - you I think) or 1/30th
sec (US/Can). The amount of movement by the subject in that time
could be large. You can decrease the "exposure" time by only capturing
one field of the video signal (two consecutive fields are interlaced to
make up one frame). That reduces the times to 1/50th and 1/60th sec.
Not good but better. The disadvantage is that your vertical resolution
is halved. The single field mode is sometimes termed "Action".
There are several ways of getting good action shots: 1. Pure luck. Sometimes you just catch the subject at the right moment. This is the #1 factor! The only remedy is to make lots of captures and hope some are good. If you had Miro DC30 card (megabucks) ratherr than the pctv, you would have the ability to capture full video (with suitable hard disk). You could then do continuous capture, at least for the action sequences. Then you could review the frames later to get the best. The cheap alternative to recording full video is to use a standard video card with high speed hard disk and record the show as an avi. There are various codecs for doing this - you can probably do it with your existing cards. You can go through after recording and select the pics you want. It does require very large amounts of memory. If you only require still pics then you could record at 2-5 frames per sec which will cut down the memory and still give good choice of frames. If this sounds interesting you could get in touch with Spencer D or Legman who use the method. 2. If the tv camera is panning across the subject, you can just about guarantee a poor pic. In action photography you can reduce blur by panning your camera with the subject. With a tv you have no control, your only hope is if.the tv shot pans with the subject. 3. If you capture the subject moving towards or moving away the relative motion is reduced so more chance of a good pic. Left-to-right action is bound to be high movement. 4. Many motions do in fact have a still frame eg the end of a sword swing, the apex of someone jumping up. Try and catch this moment. After a while you learn to hold off clicking until then. If someone is getting on a horse, you either click just as the foot goes in the stirrup, just as they have raised themselves up, just as the free leg swings away across the horse, or when fully seated. Not halfway up! 5. Although it may not be a problem for all, if you have digital tv and the signal is overcompressed, any sudden picture changes (often associated with action scenes) will cause the picture to pixelate. If this happens, you have to wait a fraction of a second and hope the signal catches up with the action. 6. Motion blur is not all bad. A swinging (blurred) sword conveys a sense of motion. A stationary and moving subject can convey an interaction between the two as in a punch just delivered. Hope I got everything and this is of some help. I think you will find any regular contributors will give you specific advice if you ask. Starman. |
Subject: Re: screenshots ?
Date: 05 Mar 2001 12:30:14 GMT
From: callistoee@aol.common (JasonL)
Newsgroups: alt.binaries.multimedia.xena-herc
| Chacha,
Just to add to the subject of video capture as one means of getting better caps: It is, in fact, *possible* to capture video with the PCTV at full resolution and full frame rate (With NTSC, that's 640x480 at 29.97 fps), should you wish to. I've recently done so successfully with a free program called Virtualdub and a free codec called Huffyuv and am pleased with the results so far. *However*, unless you have a fairly fast hard drive, you may not be able to consistently manage a full frame rate. (The CPU needs to be probably around 500MHz. It varies, depending on the rest of the system.) I used a single 80GB Maxtor drive, which is faster than is necessary, but just *how much* faster is hard to say. What happens, though, when your hard drive (or your CPU) can't keep up, is that you merely drop a few frames randomly here and there. IMO, even if you have a very slow system and can only manage an average of like 5 fps, that must surely be better than just clicking when you see something you want to save and hoping you get a cap that isn't too far after it. It would drive me crazy to continually see good cap candidates and just miss them. Nonetheless, I must say I had a fair amount of trouble getting all this working, so bear that in mind. Below are some links for future reference. (But if you have difficulty with English, perhaps you might find better venues in another language.) Below that is a very technical discussion involving frames and fields that would be difficult for anyone who hasn't actually done video capture to follow, but Starman's excellent post got me in a technical mindset and it might be of some interest to anyone weighing the benefits of capturing video: Virtualdub: http://www.geocities.com/virtualdub/index.html Huffyuv (Be sure to use the "Predict Left" setting for full res, full
frame-rate capture or
Matrox Destop Video Forum (Use the search engine to search past posts.):
http://forums.matroxusers.com/cgi-bin/forumdisplay.cgi?action=topics&forum=
VCD Help Forum (Again, use the search engine for past posts.): http://www.vcdhelp.com/phpBB/viewforum.php?forum=2&1762 Google's Search Engine of Deja's Usenet Archive (Many helpful posts
scattered about in the rec.video.desktop group): http://groups.google.com/
Starman wrote: You can decrease the "exposure" time by only capturing one field of the video signal (two consecutive fields are interlaced to make up one frame). That reduces the times to 1/50th and 1/60th sec. Not good but better. The disadvantage is that your vertical resolution is halved. The single field mode is sometimes termed "Action". IIRC, all capture cards automatically combine every 2 fields into 1 frame in hardware. And simply deinterlacing a frame containing interlacing artifacts looks better than I thought it would. (Deinterlacing may be done automatically by your capture software, but if you're doing it in Photoshop, be sure to try both the Even & Odd settings of Deinterlace, as they each will give you a different moment in time--a difference of 1/50th or 1/60th sec. Since I use NTSC, from this point, I'll just refer to 1/60th sec., rather than 1/50th.) Just in terms of resolution, ideally you'd want to have two consecutive fields that, when 'woven' together, give you a frame that is free of interlacing artifacts--that is, 2/60th of a sec. during which nothing onscreen moved. If you can find such a moment, the total resolution of your capture will double. (Double either the width or the height of a rectangle and its area will double.) That means twice as much 'real' resolution. (When you use a Deinterlace filter, your imaging program is just making up imaginary pixels to fill in blank spaces--blank spaces that make up fully *half* of a frame). This is the best of all possible worlds. Unfortunately, this doesn't always happen. <g> |
Next best is if, say, for example, Xena freezes
in place for 2/60th sec. at the apex of a sword swing, but Gabrielle OTOH
is swinging her staff full speed during that time. Gabrielle is in
the distant background, while Xena is practically swinging at the cameraman.
Xena looks great in this frame, but Gabrielle is horribly interlaced, looking
more like a small pile of jagged lines than a person. Well, since
she's off in the distant background, Gabrielle is so small that she wouldn't
look good even if she were motionless. You just want that nice image
of Xena in the foreground. Still, that interlaced staff-swinging
Gabby in the background is distracting you from Xena's violent beauty.
You could spend a long time cloning her out or you could simply paste part
of a deinterlaced copy of the image over her.
The next worst scenario requires a more detailed setup. This is where the benefits of video capture start to come into play. Say in frame 1 that a stone-faced Xena is finishing blinking. In frame 2 her blink has finished, meaning we know her eyes are motionless for the 2/60th of a sec. that constitute that frame's duration. But blast! While Xena's eyes aren't doing anything in frame 2, her mouth happens to choose that 2/60th sec. to open and begin speaking, an action that naturally will continue into frame 3, frame 4, frame 5, and god knows how many more frames. If not for Xena's mouth being interlaced, frame 2 is just the image you've been searching for for hours, so how do you salvage it? Simple: Cut out the eyes from frame 2 and paste them onto frame 1. Ah, but what if, in frame 1, Xena's mouth *hadn't* been motionless? What if, before frame 1 even began, she'd been yawning and didn't finish yawning until halfway through the 2/60th sec time period that frame 1 covers? Then her mouth would appear interlaced in frame 1 as well as frame 2, so the above solution fails. However, you can take advantage of the fact that while her mouth appears interlaced in both frame 1 & frame 2, in reality it *was* motionless for 2/60th sec. That 2/60th sec. is composed of the 2nd 1/60th sec of frame 1 and the 1st 1/60th sec. of frame 2. You see, the order in which you 'weave' fields together is arbitrary. For a given field, you can 'weave' that field together with either the preceding field or the succeeding one. Frame 1, frame 2, and frame 3 are actually 6 fields. Let's use the notation Frame 1 is composed of field 1 and field 2.
Thus, to get an image where both Xena's eyes and her mouth are motionless, you'd combine field 2 with field 3. You can do this manually with a mask (Make one mask with all the odd rows black and another with all the even rows black.) or there's a program called Avisynth (made by the same person who made Huffyuv) you can use that will play back video like so: NewFrame 1 is composed of field 1 and field 2.
Thus, using Avisynth, all the interlacing of Xena's eyes and mouth in
this case would be no problem at all for you. You'd simply save NewFrame
2 as a file and be done with it. Avisynth is a free plug-in for VirtualDub,
but it takes some time to learn.
Jason L |
Subject: Re: screencapping help needed.
Date: Sun, 24 Jun 2001 12:03:00 GMT
From: "Jason L" <callistoee@_____.com (yahoo)>
Newsgroups: alt.binaries.multimedia.xena-herc
| The USB version of the WinTV has to use a lot
of compression to squeeze video into a megabyte per second. You should
expect to see all manner of distortion when watching it on your computer,
especially when the source is tape. But when you take a screen cap
it should look better.
Just to clarify, to get the best quality you have to select the option "2(Still)." It may seem counterintuitive to use the "Still" setting if you're capturing things that are moving, but it will give better quality caps for both shows X:WP and H:TLJ. That's *if* the frame you happen to capture isn't interlaced. If it is, then you can just deinterlace it yourself with the methods Mike's covered before. |
You'll find that even things that are moving
won't always appear interlaced. In fact, no more than 40% of an episode's
frames appear interlaced (due to a process called telecining), and that's
with an NTSC signal. There's even less interlacing with PAL (sometimes
none, depending on the broadcast). Thus, more often than not, the
frames won't need deinterlacing and you'll get twice the resolution you
would if you were using the "1(Motion)" setting, which deinterlaces every
single frame you capture.
--
|
Back to Basics To main Hints 'n Tips page Forward to Graphics Software
Or, you could go back, to the main Xena page. Got something to say? Whyn't you write? webmail@dancingbones.org
