People say...


A few excerpts from E-mails I got from people speaking of their impressions about SpotRemover.

It's hard to say how many kids in the past few years have enjoyed the nice and clean filtered rips of cartoons and family movies. I will say only that during this time, I have made about 200 rips with SpotRemover for our movie archive, the last ones were made for this Christmas.
So, your work and efforts were (and are) very helpful and became a real benefit to many people.
Thank you!

1. You say this filter is slow? No, I would not say so.
There are filters that are much slower. Sometimes, when my material needs serious processing, the one pass ripping takes about 2 days. That is 4 days for a 2-pass rip. So what? I am not in a hurry. I take it like this: I spend my time just once, even though it's 2-3-4 days. But then, for years later, thousands of downloaders (and that's about the scale of our website) will enjoy the rip that was made with best possible processing standards.
Just think: a week later, no one except the ripper itself will care about how many hours were spent on this rip. But the quality of his work will definitely be appreciated and thanked.
That is my position. I am always ready to take responsibility for any work I did.

2. About the free competitors.
Before I chose SpotRemover, I have too researched a huge number of free filters.
And I can tell you: this filter has no rivals.
Take any fast scene (where you have a noticeable motion) and with fine details. For example, riding by a forest, where the camera is panning and there are small leaves. And do your comparison on these fast scenes.
I have carried out such an experiment, so I can tell you my conclusion.
ALL filters produce artefacts. ALL of them. And this one too.
Because a filter has to resolve an impossible compromise: should it replace a block, presumably a spot, with a copy from the previous/next frame? And this has to be resolved in motion, when the previous/next frame has obviously no appropriate block at that place.
But! SpotRemover produces by a magnitude fewer of such artefacts, which, besides, are much less noticeable. And SpotRemover, unlike other filters, is very flexible in fine tuning. And what is more important, it can be customized very efficiently with its scripts. (I am using those, they are a real power!)

Dimon (administrator of arjlover.net, the author of MailDownloader)
January 08, 2011.

My comment:
It looks like Dimon is using SpotRemover without AviSynth+MVtools.
From my experience, MVtools allows SpotRemover to virtually get rid of the artefacts completely.

Thank you for your EXTREMELY FAST response! I have installed SpotRemover and it works very well.
An AVI video file with 31 minutes and 30 seconds length was rendered through your plugin in about 37 minutes (approximately 22 PAL frames per second of this color 720x576 pixel interlaced material, only video - no audio stream). I used VirtualDub version 1.9.2 on a Quad Core Intel machine (2.66 GHz) with 3,25 GB useable RAM under 32 Bit OS "Windows XP Home Service Pack 3". It was rendered from a SATA II RAID (level 0) source to a destination SATA II RAID (level 10).
SpotRemover was able to remove relatively large film scratches, spots, blotches and it reduced flicker (luminance changes "pulsating brightness") which resulted in a much cleaner and more stable "easier to watch" video. The original film was delivered to me on an analogue Betacam SP cassette, so you can imagine: film problems plus analogue video streaks (horizontally) and noise and irregularities ("washed out" colour regions) in colour reproduction... *But SpotRemover helped a lot to clean up this mess!
*You are doing a fine job, thanx again.

Best regards
Ernst

Upon my comparison of the noise reduction capability of Neat and SpotRemover, I can't see any difference in quality except that SpotRemover is considerably faster. I wish I had been aware of SpotRemover earlier. I could have saved myself the $99.00. It was just chance that I came to compare the two products. You could check this out for youself and then let other's know of SpotRemover's capability compared to Neat's.

Regards,

Ralph

I finally got Spot Remover working with AviSynth script and your are right, I'm amazed with the results. There doesn't appear to be a need for me to experiment with any more Spot Remover settings as the ones included on your web site seem to do such a good job. Even hair line flaws were repaired fully...
I want to thank you again for all your help. It's unusual to find such good support for anything these days.

S. Clarke

Thank you for a very useful filter - it has proven to be be very good at cleaning up the childrens video tapes for conversion to DVD.

J. Wright

You made an incredible filter.
I am using VirtualDub 1.5.4 and the results from SpotRemover 3.5.3 are literally amazing.
Thank you very much for all your hard work and good luck on future projects!

J. Doust

This is a SUPER product. Thanks so much for your work. I do lots of 8mm and 16mm tranfer, and your demo was fantastic to say the least.

Richard

I indeed like your filter. Good job.
I little bit less liked limitation for frame spots number for processing. On most of the cases, time is not as imortant as quality ...
Thank you again.

MK

... well, of course I found out about your filter only today, after having encoded all three (quite spotted) Star Wars original bootleg DVDs and just being in the process of encoding Indy I (from the same Malayan source, and just as speckled...). These spots have always annoyed me very much, and just adding a spatial smoother didn't really cut it.

You can't believe how angry I was at Lucas after seeing that his oh-so-digitally-remastered version he most definitely had a whole company sitting at still had so many glitches... This really is a shame. As if the fact that the DVDs were just Laser Disc transfers and hence a little worse than a real DVD wasn't bad enough :-( ...

First, pixel locking is so daaamn cool :-))) ! It really seems to make the movie much more stable (seems, because, although I have no doubt it really does, I achieved nearly (!) the same results with SmarthSmoother HiQ - in this particular aspect, I mean -, but I'll refer to this later). Just tried one short scene with it, and it looked simply great.

SmartSmoother could have done this, too - but: I'd use a diameter of 7 to achieve almost the same smoothness/detail factor (slooow!), and it'd still kill some minor details. Plus, it surely isn't as exact, as it's 2D, as you said yourself, so if the same pixel doesn't change chroma/luma in time, that's merely a coincedence (whose probability can be increased by using big and hence slow values, though).

The debug locking (or whatever) feature is great, too. First I thought "Heck, nice work, but obviously totally useless for the average Joe, as nobody sees a damn thing here!", but slowly I got the idea - mainly motivated via your "will help finding the optimum smoothing/locking values", which I really wanted, of course. So, basically, I'll try to find locking values that make all real areas (let's say downright black shadows) totally clean (= bright pink, not dark magenta, not "almost white"), right? This really looked awesome later!

Then, actually a triviality, but nice nonetheless: I like the brightness/contrast/saturation stuff. Hope I'll never need the blue/red slider, though, and will keep my fingers from saturation, too, if possible. But seeing as you allow 1-percent steps (as opposed to the internal filter only allowing for 6% in|decreases), I could finetune the picture a little better. Now I finally got some 0 and some FF values in the pictures, as opposed to the original (unless I boosted the contrast with the old contrast filter to some whopping 18%, which certainly resulted in clipping). Fine, too :-) !

Speed is more than just okay (Seeing from my test encode yesterday, it doesn't really change that much compared to my former approaches with SmartSmoother HiQ, Diameter 7 - now I do still use that filter, but only with Diameter 3 or 5...), so it will still not take me more than just 48 hours for a complete 2 hours, 2:35.1 movie. I can perfectly live with that.

And the quality - wow! The pictures look sooo clean now :-) ! In fact, if I look at the Uncle Wang release of Indy and compare it to what I got out of it, it is a whole new movie :-) ! If only these Malaysian freaks had spent this time!


So, I think I'm finally getting a grasp at finetuning the filter. My values still are ludicrously low, but I'm not worried about that: Your defaults obviously are for good, but _a bit_ speckled movies, while my testing candidate is absolutely soiled in some scenes (probably stock footage - actually a good situation for scripting ;-) ...).

I finally found the settings so incredibly good that they removed virtually any real spot in the intro (the sky scene), and that one was very, very soiled. It just looked perfect afterwards!

Personally, I like the rectangle idea very much, and it helps a whopping lot tweaking the spot detection. My absolute dream would be coloured markings for failed the spot clearance, failed the perimeter clearance", and spot too thin, so as to give the user a hint of which _exact_ parameter to tune in order to include that particular spot as well, but I guess that's a) not possible (you probably process the picture data at a later point of time where these ex-spots have already been discarded from the tables), and b) not really feasible, as the whole picture would be full of distracting rectangulars ;-) . Still, the rectangles are very helpful the way they are already.

Besides, I have to emphasize that again. I have now achieved scenes that are so clean and stable, they don't look significantly worse than my Star Wars Episode II DVD - and that really means something! You know this fuzzy warm feeling you get when you see a clouded sky, and the blue is a deep, solid, unartefacted area, with the clouds standing out cleanly and all that? And on the ground, there's a grassy plain where you do still see each straw (Well... think of the resolution, but anyway), but without all those pixels wobbling around? Just great!

So far, I could only achieve this with a heavy dose of smoothing and later sharpening the stuff again. Needless to say, the grassy area just looked like a green pond after that ;-) . Not anymore, no. Now it just looks perfect!

Meanwhile I have to say that your defaults are indeed the best all-purpose settings. They don't help me in my particular case (Heck, maybe I should delete the Indy VOBs and try another movie for a change - er, like Indy II ;-))) ?), though, as they're much too tolerant, and the intro is waaay too speckled. There's really no chance to get rid of these speckles except for setting your 26/18 to some incredibly brutal 11/0. Sorry to say, as this - while making the intro really DVD-like - would of course screw up the complete rest of the movie...

So, what I did - just for testing - overnight was letting the movie calculate with only locking enabled. This already looked great, as I expected. By the way, my estimation was correct - your filter hasn't slowed down anything very much (especially seeing as I could decrease the SmartSmoother diameter significantly!).

Oh, and the quality gain was of course also reflected in the size of the file (compressability): Same encoder settings for DivX, but one time my old perfect settings, one time the perfect settings with your filter:

169.964 KB versus only 137.082 KB :-))) !

That is, the movie's size has shrunk to only 80 percent of its former size!

Thanks for all your great work!


Finally, I did it the real way, i. e. like I would do "in the wild". You've convinced me of using the script stuff indeed. So I first encoded the whole damn movie with my (= most of them being yours) default settings, then checked the whole thing and took notes as of which scenes to mark still and which to mark high motion. Three plain still scenes, and, um, about 45 scene-fragments where blocks were created, so I had to set them to high-motion. This - the checking - took me some 4-5 hours. No kidding. But it was well worth it.

The result? Well... Let me put it this way: Normally, I'm not really a friend of rude language, and usually, I don't speak to myself aloud, and especially not in other languages than my native tongue. Still, when I saw the output (= preview), even without DivX postprocessing, the only thing that came over my lips was unf••kingbelievable ;-) ! Aw, this looked sooo damn cool, I still don't believe it! I guess that with an even bigger harddisk (for storing the way bigger HuffYUV files) and a way more potent system, it may be worth the while resizing the whole sucker to 720 * 270, add a black border so as to make it 720 * 480, and do a real DVD version for sending back to Uncle Wang, with a small letter like "You do see the difference, idiots?" ;-)

And hey, forget about my much too modest praise of the space saved! Have I really been talking about 20 % (= output file only 80 % as big as with the former, pre-SpotRemoval settings)?

I was lying.

Want some actual figures, calculated for a whole 1:52 hours movie? Here they are:

1.416.730.624 bytes for the earlier version (no sound) and: 0.928.266.240 bytes for the new one (preview)!!!

Speaking of improving the compressibility, eh?

That means, it has been shrunk to less than two third of the former size! Not just 80 %!

Geez, I'm sure this _will_ show when I have to decrease the quality (bitrate) while aiming at exactly 700 MB (including sound). If that's not a big improvement worth the wait, I don't know what is...

Anyway, I thought my rips were already as perfect as they could get with 700 MB (and in fact they were - compared to what's available on the net, that is), but now I see that there was still room for improvements. 34.6 % room, to be precise ;-))) .

That's great! I would have expected a lot more glitches and errors, but the way I see it, the filter itself is now almost perfect. I wonder how much more perfectly it will behave if dealing with real DVD stuff instead of these crappy Malaysian bootlegs. But if it could handle the latter, it will probably swallow everything ;-) .

Chris





Main page




Updated: 09/22/2011