try NeatVideo it does a great job, not free but worth every penny... IMHO.

+ 1 for Neat Video. Not free, but it's GOOD.

Regards,

Kimberly

+1 For Neat Video. Has saved my bacon on many occasions.

Neatvideo works great with analog source.
Can someone show me some comparisons with digital source?

Thanks to everyone for suggestions, I am experimenting with Handbrake, very nice utility. Not as slow as I was expecting, i7-4930K seems to munch through video files reasonably quickly and the batch processing works well.

Hopefully we can avoid the need for this in future with a Sony A7s ;)

digifish

Please don't expect any connection between Sony HW and SW. Two
different divisions.

>Please don't expect any connection between Sony HW and SW. Two different divisions.

Not expecting anything but stellar high ISO performance



digifish

Possibly stu-pid question but why are you pointing a camera at the screen rather than capturing it?

If you cannot get any of the screen capture programs to work well with the application then an Edirol VC-300HD is fool proof but you do need it plus a recorder.

Bob.

I use Replay Media Catcher to capture YouTube. Sometimes several versions at different qualities get downloaded, so I save the version with best quality, without transcoding (usually MP4). Not only do you end up with better quality due to no recoding, but you avoid freezing or glitches if the download bit rate is on the slow side.

If the video is copy protected flash, like the TV video on demand channels, then the best option is to use copy software that records a defined area of the screen. The video is recoded, but you avoid the issues with incorrect colour balance of your screen, etc. that you would get by shooting a computer screen.

Sorry, I think I misunderstood you.

Is it the instrument itself being controlled that you want to capture, so that you can put it on YouTube etc?

Bob >Possibly stu-pid question but why are you pointing a camera at the screen rather than capturing it?

...for artistic reasons.

We normally use Camtasia for captures.



digifish

Just thought I'd add this video about two different De-Noising Programs:


I have several video clips that suffer from too much noise, so hopefully I can find something to help my vids out as well.

It seems like Canon camcorders are bad about giving you grainy images, and work poorly in low light. That's the case with my Canon HV-20 and my old Canon ZR-100.

I am in the minority (possibly the only person in this forum) who does not care for Neat Video. Why? Well, I am from the school that thinks "noise reduction" is just that, a reduction in noise, rather than total elimination.

But wait, wouldn't it be great if we could eliminate all the noise?

Yes it would, and I think most users of Neat end up choosing settings which do exactly that, and love it because all the noise is gone.

However, in my experience, so too is a lot of the detail.

The real art in noise reduction comes in learning how to balance a reasonable reduction in noise with both the inevitable elimination in detail, and also the introduction of various forms of new artifacts, many of which can be more distracting than the noise itself.

Video noise, like film grain, is something we've grown accustomed to, like the hiss on an analog tape, or minor ticks on even the first play of a new LP vinyl record. Because we are accustomed to it, it doesn't bother us as much as new artifacts that we might introduce.

I have suggested on several occasions that someone post a problem video, and everyone take a crack at doing noise reduction. The folks over at doom9.org do this fairly frequently, and there are all manner of noise reduction techniques discussed in that forum. Here in this forum, I don't know if I've ever seen a before/after using Neat Video. The closest we came to that, that I remember, was when Nick Hope posted some of his underwater video and we all tried to denoise it for him:

My Latest Denoising Tests (thanks Nick!)

That link actually goes to a post that I made, but I provided the links to Nick's original posts, as well as to the discussions in the doom9.org forum.

Bottom line: for me, I have not found anything that comes close to what I can do with several AVISynth plugins. I just spent several hours yesterday further refining what I've learned to do over the past ten years, and found ways that actually leave a little more noise, but preserve far more detail, not just the fine detail, like hair, but details in fairly flat surfaces, which often get "mushed" together.

If anyone wants to post 10-15 seconds of original, unprocessed video that they would like de-noised, I'd be happy to take a crack at it. Hopefully someone who has Neat can do the same, and we can let everyone view it on their big-screen TVs and see what they prefer. I wouldn't be surprised if the majority still favor the Neat video result, but if you look closely, I think you'll find a lot of details are gone.

I think most of the fish video before/after videos are still available in that link I provided above, so for SD video you can get a sense of what happens with various de-noising techniques.

John, are you saying NEAT is not adjustable at all? Can't change the amount of NR?

Yes, you can adjust it and it does an excellent job as long as you know what you are doing. However, and I think that is what John means, you can easily overdo it and end up with a video that looks very "plastic" almost cartoon like. Already the standard setting which is around 60% is way to high; I usually go down as low as 30% and sometimes lower depending on the scene.

I use Neat Video frequently, and it is a plugin to be very careful in not overdoing. I always examine the noisy clip with a color correction plugin before even attempting a denoise, because a lot can be done to minimize noise there.

Chroma noise is much more objectionable than luminance noise, and if you have a color corrector similar to NewBlue's Colorfast, you can reduce saturation in heavy shadows and get a much better look immediately. I see what improvement can be made with color/gamma/saturation correction and that gives me an idea of how lightly I can use Neat on the clip to get some degree of improvement before I apply final color correction.

I totally agree that "reduction" is the important word to keep in mind. Total elimination of noise with Neat can leave your clip with detail approaching a VHS tape.

Yes, Neat lets you adjust the amount of noise reduction.

My objection to Neat is that is supposed to be a "temporal-spatial" denoiser. This means it denoises using two different approaches simultaneously.

Spatial denoising looks at each frame all by itself, then examines adjacent pixels and, in the case of Neat, uses a noise print from a featureless frame (something you supply) to identify anomalies it thinks are noise. It then replaces these noise pixel groups (noise is more than one pixel!) with some intelligent blurring, additions, and subtractions from adjacent pixels.

Temporal denoising looks at the two adjacent frames (before and after) and looks to see what pixels have changed, and uses that approach to decide which pixels are noise. To make this work really well, the adjacent frames are first "motion compensated," meaning that they are moved, stretched, and adjusted to make them "perfectly" overlay the current frame. If this motion estimation is done well the algorithms can just subtract the two images. What remains is noise, and you can then blur, add, subtract, etc. to make those noise pixels disappear.

Temporal denoising is incredibly complex, but has the advantage that it can retain a huge amount of detail. It's big disadvantage is that is can result in all sorts of strange residual artifacts when the camera pans. The result sometimes looks like it was filmed through a screen door.

Spatial denoising generally does a better job of removing noise, but it always cuts down on detail.

I have always felt that Neat relies way too much on spatial denoising. If you download some of the "fish" denoising samples in that old thread I linked to, you'll see the Neat sample that I provided. You'll see absolutely no noise whatsoever, but you'll also see that the poor fish no longer have any fine lines in their dorsal fins, and many other details are lost.

If you further study those fish samples you'll find one other miraculous thing that can happen with really well-done denoising: new details emerge that were hidden by the noise! This happens because temporal denoising uses information from multiple frames to reconstruct the video for the noise pixels that are being adjusted. That averaging across frames allows subtle details to be discovered.

You know you are doing a good job of denoising when you can distinguish more detail in the final video.

I think I may still have some HD denoising I did for a member of this forum a few months ago. It was horribly under-exposed, and when brightened (using Color Curves) the sensor noise was so horrible as to render the clip unusable. I can post this clip, which I set to use a before/after show, if anyone is interested. Because of the under-exposure, the resulting brightened & denoised video doesn't look like pristine well-exposed video, but it does look pretty good, like video shot at dusk should look.

I am not at my PC right now but I believe NeatVideo has a setting to control how many frames are used for temporal denoise; the more frames selected the slower it becomes.

John, here is some footage you might be interested in. In the full film, there is some footage that has some planes flying in formation where no other footage of these planes has been found. Film from Tripoli, right before or very early into the Korean War. My friend is trying to get the original film from his Dad, as I suspect the original transfer to VHS was not done ideally?

A museum is interested in the footage, so I'd like to run it by you. Truly awful quality, would love to see what you can do.

http://www.visualretreat.com/vegas/temp-sampleForSony.avi

+1 to @musicvid10 suggestion in Handbrake filter. Given the price it does a really nice job.

I'll look at the video.

Everything starts with doing a good transfer. Anything transferred to VHS will be compromised, not only because of the low resolution inherent in VHS, but also the fact that most transfer houses did a terrible job of creating telecine (duplicate fields) in order to get the 16 or 18 fps film to play at 29.97 fps.

I'm looking at the film right now, separated into even and odd fields so I can see how it was transferred.

It looks to me like a transfer done with an old Goko TC-20. I owned one of these for about five days, back in the early 2000s, but immediately sold it back after I discovered that it blurred the image something awful. It used a rotating mirror to blend adjacent frames together in order to eliminate the flicker you get when pointing a video camera at a normal projector with a shutter that opens and closes several times for each frame of film. Since the Goko had no shutter, the exposure was constant from frame to frame. That was the good news. The bad new is that the machine created blended fields, which made the transfer look really blurry.

Here are two frames from your footage, separated into fields, with each field duplicated in order to resize the image back up to 720x480 (i.e., it was bobbed).



You'll see that the second field of the first frame is a blend of the first frame of film with the next frame of film. This is what makes the image look so blurry.

Someone over at doom9.org got some help from one of the doom9 gurus ("Didée") and was able to create a script that can sometimes recover just the unblended fields, and then put these together to get a more-or-less "frame accurate" transfer. This is no substitute for using the right equipment in the first place, and this is most definitely what you should do if you think you can still get the original film. You'll save yourself a lot of headaches, and the result will be infinitely better.

However, using that script, it might be possible to remove most of the blends, and get something approximating a frame-accurate transfer. A frame accurate transfer means each film frame gets transferred onto exactly one frame of video, with no duplicates, no drops, no blends, and no telecine (repeated fields). Once you have this, then you can start applying some of the "magic" scripts which stabilize the film, remove dirt, reduce grain, etc.