Why is Neat video the best video denoiser?

[Truthler]

Shooting underexposed in the dark, and later brightening the footage is what actually makes noise. A nice, small noise challenging denoisers around the world to keep as much detail as possible

But as you can see, nobody could challenge the Neat video in this forum.

[rrats]

Shooting underexposed in the dark, and later brightening the footage is what actually makes noise. A nice, small noise challenging denoisers around the world to keep as much detail as possible

But as you can see, nobody could challenge the Neat video in this forum.

1. they are expecting you to do it. nobody would pay 100$ to win in internet argument.

2. avisynth is way better at denoising imo

[rrats]

Also higher ISO value just makes noise bigger.

[Truthler]

Shooting underexposed in the dark, and later brightening the footage is what actually makes noise. A nice, small noise challenging denoisers around the world to keep as much detail as possible

But as you can see, nobody could challenge the Neat video in this forum.

1. they are expecting you to do it. nobody would pay 100$ to win in internet argument.

2. avisynth is way better at denoising imo

#1 because nobody admire the forum members enough. Take an example of me. I bought two Ax700, one in 2018 and an other in 2020, and electric gimbals (a DJI and a Zhiyun) for them. I bought a new laptop to record lossless NVENC and two Camlink 4 K for the cameras. I also spend money for A P. Pro and Neat video. I bought DJI pocket 2 camera too. Why? Just to show it off for you dear videohelp members. I rarely used my equipments, I never take them to my holydays. They were bought only to post things on Videohelp. It was the only real purpose.

#2 I did not see any proof for that, only empty promises.

[rrats]

#2 i did not see any proof for that, only empty promises.

because you wont send us the bloody clean looking source, export and step by step process along with the denoiser values

[Selur]

as a side note: I gave up on that file sharing service, for some reason the connection to me to that service is horrible. Slow download, upload aborts 2 or 3 times after xy gb.

[Truthler]

as a side note: I gave up on that file sharing service, for some reason the connection to me to that service is horrible. Slow download, upload aborts 2 or 3 times after xy gb.

I can not understand the situation. It works like charm, the uploads are fast even by the 2Gbit/s optical internet standards...

It is a fantastic service, I registered there with 10 e-mail addresses, so I can upload and keep 10Gb files on each account. 10X10GB

Do you have problems with the quality of internet connection?

[Selur]

Do you have problems with the quality of internet connection?

Not really, got 100MBit download and 30-40MBit upload when I'm at home and they usually work fine.

[sophisticles]

#1 because nobody admire the forum members enough. Take an example of me. I bought two Ax700, one in 2018 and an other in 2020, and electric gimbals (a DJI and a Zhiyun) for them. I bought a new laptop to record lossless NVENC and two Camlink 4 K for the cameras. I also spend money for A P. Pro and Neat video. I bought DJI pocket 2 camera too. Why? Just to show it off for you dear videohelp members. I rarely used my equipments, I never take them to my holydays. They were bought only to post things on Videohelp. It was the only real purpose.

This is just sad, and even sadder if true.

[lordsmurf]

I know there are a lot of people who wear glasses for correct eyesight , but they did noit wear it on the streets for aesthetic reasons.

Contacts.

In fact, contacts are often better, those don't warp vision. Glasses warp.

I also noticed that 90-95% of the people who deal with electronics, computers (videoediting programming etc..) during their worktime usually had to wear glasses. I can not prove that,

The technical term for this is "pulled from ass statistic".

but it has a high chance that Cornucopia, Selur or Poisondeathray use thick eye glasses during their work. LITERALLY their eyes are victims of their profession.

Thick? Very small % likelihood, actually.
But wearing something, anything? Sure. Almost all people need correction to reach the mythical "20/20" (6/6) value -- which in itself my be a flawed methodology for vision testing. The main reason being that so much screen time is not natural -- yet. Evolution (that thing some people don't believe in) may have something to say here, for future generations.

a well known social phenomenon among tech guys.

This is a complete BS statement.

What's going to happen is people will shoot, hopefully with the proper camera settings, something you show no desire to do, then if there is the need to denoise, adjust brightness, adjust contrast, sharpen or any other adjustment, they will do it

If the shot isn't good, it's usually just dumped and re-shot. They don't screw around with the bad clip in salvage/restore type programs. Salvage is what consumers do (and/or on consumer sources), for their one-time events, shot on cheap cameras (or expensive cameras used badly). I've worked with lots of documentary filmmakers, where my job is to attempt to salvage older videotape content, hopefully to some degree that match and fits in well with the new footage.

How else can I generate noise with a camera? MAX ISO, max Shutter, and minimal apperture...

Did you really just ask a complete camera newbie question?
ISO noise is not the same as underexposure noise. You cannot conflate those.

Do you have an AX700

Is that all you have? That's really nothing more than an expensive consumer camcorder. I'm sure it can shoot fine, but it has many limitations.

By comparison, I'm currently considering a Nikon Z6 II with a Ninja, and at least one stepper lens. (I already own the Deity mic.) I've had a recent need to shoot more, and the only limitation I'll have is myself, not the camera. (The only thing holding me back is seeing if the firmware rumors are true, for adding RT tracking. Still nothing. It may overlap with the upcoming Z9 reveal.)

I already won the contest:

WTF?

You need help. And I don't mean video help.
You're bat shit crazy.
And for that reason ... I'm out.

[Truthler]

I know there are a lot of people who wear glasses for correct eyesight , but they did noit wear it on the streets for aesthetic reasons.

Contacts.

In fact, contacts are often better, those don't warp vision. Glasses warp.

I also noticed that 90-95% of the people who deal with electronics, computers (videoediting programming etc..) during their worktime usually had to wear glasses. I can not prove that,

The technical term for this is "pulled from ass statistic".

but it has a high chance that Cornucopia, Selur or Poisondeathray use thick eye glasses during their work. LITERALLY their eyes are victims of their profession.

Thick? Very small % likelihood, actually.
But wearing something, anything? Sure. Almost all people need correction to reach the mythical "20/20" (6/6) value -- which in itself my be a flawed methodology for vision testing. The main reason being that so much screen time is not natural -- yet. Evolution (that thing some people don't believe in) may have something to say here, for future generations.

a well known social phenomenon among tech guys.

This is a complete BS statement.

What's going to happen is people will shoot, hopefully with the proper camera settings, something you show no desire to do, then if there is the need to denoise, adjust brightness, adjust contrast, sharpen or any other adjustment, they will do it

If the shot isn't good, it's usually just dumped and re-shot. They don't screw around with the bad clip in salvage/restore type programs. Salvage is what consumers do (and/or on consumer sources), for their one-time events, shot on cheap cameras (or expensive cameras used badly). I've worked with lots of documentary filmmakers, where my job is to attempt to salvage older videotape content, hopefully to some degree that match and fits in well with the new footage.

How else can I generate noise with a camera? MAX ISO, max Shutter, and minimal apperture...

Did you really just ask a complete camera newbie question?
ISO noise is not the same as underexposure noise. You cannot conflate those.

Do you have an AX700

Is that all you have? That's really nothing more than an expensive consumer camcorder. I'm sure it can shoot fine, but it has many limitations.

By comparison, I'm currently considering a Nikon Z6 II with a Ninja, and at least one stepper lens. (I already own the Deity mic.) I've had a recent need to shoot more, and the only limitation I'll have is myself, not the camera. (The only thing holding me back is seeing if the firmware rumors are true, for adding RT tracking. Still nothing. It may overlap with the upcoming Z9 reveal.)

I already won the contest:

WTF?

You need help. And I don't mean video help.
You're bat shit crazy.
And for that reason ... I'm out.

Nikon Z6 II is not a video camera, just a still camera which can shot video.

Where did the misconception came regarding still cameras?
DSLR mirroless etc came into fashion among photographers, who often had no practice in video production.
In the old times there were always 10-25X more poeople who shot photos as professional or amateur, than the number of video makers.
When their first still cameras started to record video, they quickly discovered the magic of video making...
Due to the large numbers of photo shooters (amateurs or professionals) it bacame a big fashion.... However real videographer professionals (who didn't come from the field of photography) always supported camcorders instead of still cameras.

1. form factor is worse.

2. You need external audio mics, and external recorders for good audio, in a camcorder the audio is not problem, since they had always superior recording system, even with external microphones.

3. DSLR and mirrorless cameras don't have so high real visible resolution, See their worst results in ISO IMATEST charts.

4. You can make nice bukeh with 7-9 blade IRIS of good camcorders

5. Zoom range is much bigger

6. you don't need extra ND filter, because they are built in their lens system.

7. No overheating like simple cameras, limited time and limited filesizes...

8. Unlimited length recording of video.

9 battery replacement and batteryproblems.

10. slower focusing, can not really sustain the proper focus during the zooming process.

11. You must buy so many auxiliary lens nd audio eaquipments that it cost more than the basic camera.

12. They are more noisy, due to the smaller idividual pixel surface of their sensor. (it is physics , what you can not override.) They try to compensate it with much heavier denoising process. The simple HW denoising process of cameras always will result in less detailed less sharp video frames.




A Sony sensor will always be the best, due to it is the largest sensor developer globally, and it is tech leader regarding sensors. That's why I always choose SONY imaging products. Abd it has partnership with Zeiss, which is undeniably the best lens producer (See the lenses which are chose for space telescopes and space stuffs, it is simply N1 in optics) The result is always good when the N1 sensor producer works together with the N1 optics company)


Factually, in IMA test ISO test charts, this is still the only comsumer camera on the market in 2021, which is able to 100% complete the 4K test Chart (measured lossless way)
Only professional broadcast cameras can reach it (which has larger sensor in 4K or has 8K sensor and 8K video processing thus obviously better). As I said N1 ZEISS optics and Sony sensor. Panasonic JVC Canon consumer camcorder sensors are not close to that in ISO test charts, only their professional machines....

It tells something about the FDR-AX700....

[rrats]

Good cameraman can make even Digital8 look good

[Truthler]

Good cameraman can make even Digital8 look good

Try to make your still camera work better on IMA ISO TEST Charts, when your still camera's electonics is struggling to compensate the noise which generated due to the very small surfaced individual pixels of your sensor... It won't work...
Try to keep the focus properly durig the zooming process... In still cameras (even on the best ones) if you make a frame to frame playback , you can see that their focus can't keep up immediatelly during Zooming. And zoom range is small in still cameras
Spend 2X or 3X more money than the price of your still camera for the external audio recorder and expensive additional lens systems ) Good luck!

[dellsam34]

When I mentioned earlier that excessive quoting is an early sign of mental disorder I was right, Here you go folks, diagnosis results are in:

Take an example of me. I bought two Ax700, one in 2018 and an other in 2020, and electric gimbals (a DJI and a Zhiyun) for them. I bought a new laptop to record lossless NVENC and two Camlink 4 K for the cameras. I also spend money for A P. Pro and Neat video. I bought DJI pocket 2 camera too. Why? Just to show it off for you dear videohelp members. I rarely used my equipments, I never take them to my holydays. They were bought only to post things on Videohelp. It was the only real purpose.

[Truthler]

When I mentioned earlier that excessive quoting is an early sign of mental disorder I was right, Here you go folks, diagnosis results are in:

Take an example of me. I bought two Ax700, one in 2018 and an other in 2020, and electric gimbals (a DJI and a Zhiyun) for them. I bought a new laptop to record lossless NVENC and two Camlink 4 K for the cameras. I also spend money for A P. Pro and Neat video. I bought DJI pocket 2 camera too. Why? Just to show it off for you dear videohelp members. I rarely used my equipments, I never take them to my holydays. They were bought only to post things on Videohelp. It was the only real purpose.

Diagnosis? Which diagnosis?

[sophisticles]

When I mentioned earlier that excessive quoting is an early sign of mental disorder I was right, Here you go folks, diagnosis results are in:

Take an example of me. I bought two Ax700, one in 2018 and an other in 2020, and electric gimbals (a DJI and a Zhiyun) for them. I bought a new laptop to record lossless NVENC and two Camlink 4 K for the cameras. I also spend money for A P. Pro and Neat video. I bought DJI pocket 2 camera too. Why? Just to show it off for you dear videohelp members. I rarely used my equipments, I never take them to my holydays. They were bought only to post things on Videohelp. It was the only real purpose.

Diagnosis? Which diagnosis?

That you need help, but not with video.

[DB83]

The only other possibility with that 'boast' is that the OP was being sarcastic.


But sarcasm has no place here. Only real facts and 'Truth'

[Truthler]

When I mentioned earlier that excessive quoting is an early sign of mental disorder I was right, Here you go folks, diagnosis results are in:

Take an example of me. I bought two Ax700, one in 2018 and an other in 2020, and electric gimbals (a DJI and a Zhiyun) for them. I bought a new laptop to record lossless NVENC and two Camlink 4 K for the cameras. I also spend money for A P. Pro and Neat video. I bought DJI pocket 2 camera too. Why? Just to show it off for you dear videohelp members. I rarely used my equipments, I never take them to my holydays. They were bought only to post things on Videohelp. It was the only real purpose.

Diagnosis? Which diagnosis?

That you need help, but not with video.

Can you explain it?

[Truthler]

The only other possibility with that 'boast' is that the OP was being sarcastic.


But sarcasm has no place here. Only real facts and 'Truth'

Real truth like comment #251 ?

I just want to help the guy to make a good decision.

[DB83]

^^ Not interested. That just takes this argument of to another tangent.


It was the, if I am correct, sarcasm in #244 quoted in #254


And excessive quoting is bad forum manners - quoting total, lengthy posts when you add just a few lines of your own.

[Truthler]

Meanwhile I denosied the street video too.

The result fantastic, even smalllest noise and grain were removed from the metal objects (like cars). Finally the video is PERFECT like a mirror!!!!

The file is here: http://sendanywhe.re/5O0QHG4Z

[sophisticles]

Can you explain it?

No, unfortunately I stopped working in a Psych ward years ago so I no longer have access to the resources i used to, but I urge you to please reach out a support group or a trained professional because you are a sick person.

Get better.

[Truthler]

No, unfortunately I stopped working in a Psych ward years ago so I no longer have access to the resources i used to, but I urge you to please reach out a support group or a trained professional because you are a sick person.

Get better.

Hello user Sophisticles!

First, Let allow me to say thank you for your personal attacks

Were you patient or nurse/doctor in that hospital? If the first, I'm glad to hear that you are free again! If the second, I hope you could avoid the occupational injuries after the long service.

But, let's return to the topic. What about post #261 ?

[lordsmurf]

Where did the misconception came regarding still cameras?

You just have dated thinking. It's "digital vs. film" all over again.

DSLR mirroless etc came into fashion among photographers, who often had no practice in video production.
In the old times there were always 10-25X more poeople who shot photos as professional or amateur, than the number of video makers.
When their first still cameras started to record video, they quickly discovered the magic of video making...
Due to the large numbers of photo shooters (amateurs or professionals) it bacame a big fashion.... However real videographer professionals (who didn't come from the field of photography) always supported camcorders instead of still cameras.

I don't disagree with any of that, but it's dated thinking. It's not been true for years.

2. You need external audio mics, and external recorders for good audio, in a camcorder the audio is not problem, since they had always superior recording system, even with external microphones.

Camcorder mic suck, too.

1. form factor is worse.
3. DSLR and mirrorless cameras don't have so high real visible resolution, See their worst results in ISO IMATEST charts.
4. You can make nice bukeh with 7-9 blade IRIS of good camcorders
6. you don't need extra ND filter, because they are built in their lens system.
7. No overheating like simple cameras, limited time and limited filesizes...
8. Unlimited length recording of video.
9 battery replacement and batteryproblems.
10. slower focusing, can not really sustain the proper focus during the zooming process.
11. You must buy so many auxiliary lens nd audio eaquipments that it cost more than the basic camera.
12. They are more noisy, due to the smaller idividual pixel surface of their sensor. (it is physics , what you can not override.) They try to compensate it with much heavier denoising process. The simple HW denoising process of cameras always will result in less detailed less sharp video frames.

Most of that is a mix of preference, or model-specific issues. It's easy to compare bad camcorder with good mirrorless, or bad mirrorless with good camcorder. But 1:1 is less obvious. You specifically mentioned the AX700, which is just a costly consumer model camera. And I specifically mentioned a Nikon that is more advanced. (The Z9 may be a game-changer, which is the rumor, let's see! But it's not something I'll be buying.)

5. Zoom range is much bigger

Not really. Optics are optics, and you get distortions with zooms, photo or video.
Even in terms of distance (is that what you meant?), also no. For example, we have a 200-500mm f/4 Nikkor lens. That lens costs more than entire cameras. It's fast, quiet, and incredibly sharp. It will makes most camcorder zoom lenses @ max length look pathetic.

A Sony sensor will always be the best,

"always" .... I'm amused. Always, until it's not. Sony has often made outstanding sensors, but "best" isn't as clear cut. Canon and Nikon are no slouches, amongst others.

Abd it has partnership with Zeiss, which is undeniably the best lens producer

Zeiss makes tons of crap lenses, including junky camcorder lenses. I remember P&S cameras with "Zeiss" plastic optics. It's just a brand name. Ironically, not unlike Sony. The name "Zeiss" alone isn't impressive whatsoever.

Panasonic

I still remember when Panasonic was a mirrorless video darling. That didn't last. Video is still in an early phase, in terms of modern shooting abilities. Sharp, high ISO, high DR, etc. It may be years before there's a lull in the market again. So don't get too married to any brand or way of thinking. A reason I haven't really shot in the 2010s is because it got complicated -- in a good way, but still complicated. I can settle in with the Nikon Z6 II, as I have AF-S lenses available. For me, the time is right, need is suddenly here, prices are tolerable for a quality setup that won't limit me (I'll limit me). The AX700 would definitely limit me, and probably piss me off.

And my goal would be to shoot well, good settings, and not need something like NeatVideo (or Avisynth).

BTW, this was a more sensible conversation with you. Do more of that, less of the other.

[Truthler]

...,,,---...,,,---

You still can not come out with good results from the focusing during zooming, neither from the law of physics (smaller area of individual pixels creates more noise, what the camera tries to suppress somehow, which will result in a less sharp video frames and less details.

Let's remember to the oldest brand in optics, and its famous achievements during 19th century, 20th century and 21th century, :



And what happened in ww1.

British-German Trade Rubber And Optics during WW1
With the famous American ww1 historian Indy Neidel (2min)

https://www.youtube.com/watch?v=nhTXaIVVe54

[rrats]

Can you please, for God's sake, stop quoting entire posts

[DB83]

And trying to keep 'On Topic' might also help.


(And I certainly do not welcome PMs in this matter)

[Cornucopia]

[Off-Topic Post, but necessary]

OK.

I am going to say this clearly, succinctly and professionally.

They say that imitation is the sincerest form of flattery. Well, I suspect flattery wasn't the intention here. This happened on this site before (and perhaps this is a repeat from a long-gone troll under another guise). Regardless,

@ Truthler, you are in violation of using my personal copyrighted work (my avatar), or a derivative of it, without permission.*

I do not give permission, nor do I intend to anytime for the foreseeable future. I want the avatar taken down or changed to something that is clearly not it nor a derivative.
If you do not, I will be reporting you to the mods for copyright infringement.


Scott

*Image GIF animated by me in early 2000s, based on abstracted image hand-drawn on computer by me in mid-90's, based on image hand-drawn on paper by me in the '70s when I was a teen, based off my romantic memory of a Groucho book. I have files & papers to prove all their provenance and authenticity.

[codehound]

Wow. Whole thread needs locking. Shocking.

[poisondeathray]

But as you can see, nobody could challenge the Neat video in this forum.

Did you post an improved version on the newspaper that I missed ?



Methodology and sharpening:

Sharpening is not denoising. They are different processes (obviously)

You should turn sharpnening to zero because you're comparing denoisers, not sharpeners.

Ideally, you shouldn't add x,y,z filters and pollute the analysis. eg. If you're testing covid vaccines efficacy, you don't add 3 other vaccines , 4 other antivirals - because you can't tell what is doing what. This is basic scientific method. You control variables, such that only 1 input factor (the denoiser) is doing the work. Cause and effect.

It's ok in the after analysis to look at a sharpened version - because you're examining sharpened noise/artifacts vs. actual details, and applying sharpening can highlight the differences and amplify the noise. But you should apply the same sharpening filters, same settings to each case. This way you can more easily distinguish which has more noise/artifacts, vs. more actual details. For example , if you're using internal NV to sharpen, you can disable everything in NV except sharpen and apply that. (Remember , I use Adobe/NV too - I "play for both teams") . If you're using other filters you have to specify which, and what settings, otherwise people cannot replicate and it complicates the analysis. Someone can take the output of A,B,C,D denoisers (without sharpening) and apply the same sharpening algorithm to them (e.g. NV sharpen at 250% for high,mid,low - but you have to state or show the other settings like range , which channels, twirl down the menus) , so they are more comparable. Otherwise you have to try adjusting some other sharpening filters and strengths and algorithms to approx. match images, to try to compare on similar terms. If you're performing other pre or post manipuations, you have try to apply the same to both; e.g. stabilizing will result in different results unless you use the same prestablized input into the denoiser. For example, I uploaded a prestablized input for the newspaper. If Neat was the "best" denoiser, it should get the "best" results from that same input... Otherwise, how can you distinguish if some other filter or manipulation isn't contributing to some or most of the work? Splitting up individual processes and variables helps you learn about the strengths/weaknesses of each filter. But it's not a perfect world, we don't have endless time to test and to upload dozens of pre/post versions. Some compromise is good for me - enough to see that it's possible to produce similar results - eg. I might use slightly different sampling region in Neat - but it should produce similar results.




I'll gradually post some comments and screenshots over the next few days...but what you posted has problems, residual noise, oversharpened noise and details are not as clear as they could be, frame border and temporal ghosting issues

It's *impossible* to beat frame stacking methods on scenes that can be stabilized well (low camera motion, no rolling shutter atifacts, 1 focal plane, no moving FG objects, same lighting conditions, noise type is roughly "random") by using traditional methods (it might be possible with machine learning someday in the future, but currently not). eg. the newspaper video, or in this video, the building shot near the middle - you can't beat frame stacking approach. But if the shot has large motion, multiple depth planes, parallax, moving people/objects, etc.. that method is less effective, or you have to use workarounds - such as combination stacks with layers and compositing techniques, motion masks, roto, more complicated border fill methods. But it's a lot of work and usually not worth it for general use, unless you want the absolute best.

And there is no reason why you cannot combine methods. It's not one vs. the other. They are just tools, and each has pros/cons, and you should be using the tools in combination . eg. You might use Neat for some sections or parts of frames, and fix some of the NV problems in part of sections with other methods.