Chainik wrote:

It's just a common trend now, like 2K screens in the phones for example

Well, 2K screens on smartphones DO make some sense (if the screen is relatively big).
Say a Galaxy Note 3. That has a 2K display wich is 5,7 inches large.

The theoretical highest resolution that a human eye can see is a 2K display at a distance of 1,6 x Screen diagonale (or 4K at 0,8 x diagonale or 8K at 0,4 x diagonale)

5,7 inch x 1,6 = 9,12 Inch =  @ 23,16 centimeters eye distance to that display = 1 Pixel is the smallest object an average human eye can see. 23,15cm is still a reasonable distance. Means 2K for such a display is reasonable too. 4K however wouldnt. wink


(5 replies, posted in Using SVP)

Isnt "soap opera effect" simply the look of frame rates significantly higher than 24 FPS?
It even happens to material that is filmed in higher frame rates. For example "The Hobbit". People complained about "Soap Opera Effect" in it. Wasnt it filmed in 48FPS?

I recall people complained about "The Hobbit":

- It doesnt look like Hollywood Cinema, it looks like TV-series (Soap Operas). Explanation: Most TV Shows are US-American. Means, they are filmed at 30 FPS. Thats already higher than Cinema, wich is at 24 FPS. Also the interlacing of video works almost like some sort of frame blending, with the result that the FPS apears doubled.

- It doesnt look "professional", it looks like someone filmed it with a Home Video Camera. Explanation: Home Video Cameras are capable of shooting at 60 FPS for several years already. Even my smartphone can shoot at real 60 FPS (or Ultra HD at 30 FPS). People are used to the looks of 60FPS from their home made videos. So they think, thats the look of amateurish cheapness.

- professional filmmakers complained about: That doesnt look like a movie. It looks like beeing life at the set watching the actors life or like beeing at the theatre. Everything looks far too real, the magic is gone blabla no magical hollywood world but reality.... You start recognising that its all fake, because now you see that some things (Silicon masks and other stuff) dont move correctly, what you didnt at 24FPS. But at 60FPS all movement is crystal clear and you start seeing these things.

All those are not problems of "badly done HFR" or "interpolated HFR" (like some claim), its the problem of HFR and of the head of the people (their experience that professional movies are done at 24FPS and mothers film the first steps of their children in 60FPS, wich causes them to think: 24FPS looks = professional hollywood, 60FPS = cheap home made shit)

MAG79 wrote:

if blending frames to a screen refreshrate does actually anything
SVP and madVR frame blending do nothing if framerate = screen refresh rate.
TV does any what it can. For example my TV (Philips 40PFL5007T) does frame interpolation to 100/120 fps from standard HDMI signal (1080p 50/60fps). In other words my TV does additional framedoubling wink

Hm ok.
But woulnt it be better if SVP does the 120 FPS instead of using interpolated (artefact distorted!) frames as source material for another interpolating? Or doesnt the connection support 120FPS?

I dont own a TV. Specially not one with interpolation. I own a Laptop with 120Hz screen and a Projector with 120Hz and use "to screen refresh rate" and the OSD on the projection says: 119,9999 whatever FPS.

Just my thoughts about this? (interpolated frame as source = not ideal)

madVR does simple frameblending. It is not motion interpolation in classic terms.

Yeah. Thats why I said "Fake interpolation" before.

The question remains, if blending frames to a screen refreshrate does actually anything if SVP already does real motion interpolation to screen refresh rate.

Because we are talking here about:
SVP turns 24 into screenrefreshrate, then madVR turns screenrefreshrate into screenrefreshrate and the TVs blending option turns screenrefreshrate once more into screenrefreshrate.

As I said, if you think it works, then its fine.
Its just that I prefer to use ONE anti depression pill brand at a time and not swallow 5 different brands of it, at once, it might cause unforeseen problems beeing used together. wink
Or 7 installing 7 different viruskiller softwares on one mashine, just to make sure.

Forgove me, but 3 different frame interpolation softwares in a row (SVP, madVR and your TVs), sounds strange to me.
Specially if you have SVP settings to "Frame rate". Whats the "Blend to screenrefreshrate" suposed to do in that case? Because say its 60Hz, SVP renders 24 to 60 frames. And then the blender is suposed to blend 60 frames to 60 frames?
It would be something different if SVP would have set to: "double frame rate". Then the blender would at least have something to do by blending 48 into 60.

I fail to see how this is relevant to my question, however.

Not with the original question.  Just wondered why Frameinterpolation alone doesnt give a satisfying result and forces someone to fall back to an inferior (less effective but also much less hardware intensive) way of smoothing motions with some kind of fake-interpolation.
I personaly would have thought, stuff like madVR smooth motion or SVPs own one are just there for people whos hardware cant afford true frame interpolation or high FPS one. (say, real interpolation from 24 to 48 and then faking it (madVR or SVPs own one) from 48 to 60.

So I thought, maybe the SVP settings could be improved. smile

But what was the saying in homeopathy? "He, who heals, is right." You feel better with it. So its fine. big_smile

Is there any difference for you, if you use only madVR Smooth motion without SVP active?
What exactly do you mean with "Low Artefact Settings"? Do you use SVP at "M2" (wich is almost (Actually only 1 out of 5 frames are interpolated ones) identical to switch off SVP) or something?


(6 replies, posted in Using SVP)

Hmm, thanks for pointing me to that possibility.
Could be interesting for some things.

Though its still rather limited. If I stream video from my home PC to the SP, I should be in reach of a WLAN connection. (my mobile contract has a traffic limit of 1GB/month with 7,5Mbit, after that slowed to 64kbit). Best would be, beeing at home with my local WLAN. But if I am at home, why would I stream video to a smartphone? Well, except if its say, something short from Youtube and I am on the WC.  big_smile


(6 replies, posted in Using SVP)

Hi there!

I just got me a Samsung Galaxy Note 3 (my first ever smartphone).

Since I am totaly used to HFR Video, "normal" Video looks totaly crappy to my eyes meanwhile.
I converted some videos and watched it on the smartphone in Full HD, 60FPS.... works very nice.

But converting consumes huge amounts of time. (About 1.5 times the lengh of the converted movie for me)

I wonder if there is a way for motioninerpolation on smartphones? Yeah I know, we can be happy that it works for Laptops.  big_smile But I still wonder.

There is still some thing "in between" those.

"to small step 6-8" for example.


(9 replies, posted in Using SVP)

looks like it falsely detected it to be 3D.
Try to change it to 2D by hand.


(9 replies, posted in Using SVP)

Uh. I must admid that I just NOW realised that you are talking about video CAPTURE and not playing video from a external HDD as I initially thought you try.

The CPU load I mentioned was the windows task manager numbers. (stutter at 60% windows task manager CPU load, thousands of dropped frames after a few seconds etc)
And it was by playing 1080p MKV, no real life video capture (no epxerience with this in modern times. Last time I experimented with video capturing into the computer was like 1999 when I used an old (maybe 80s) Sony Hi8 Camera over a TV-Card in a Desktop-PC for early Video Chat.  big_smile


(9 replies, posted in Using SVP)

@Jeff: yes.


hen I dial-up the settings for higher quality - let's say by reducing the motion vectors grid

What for example? To 6px?

Its the same for me then. with 6px grid I get stutter @ 60% CPU usuage and 30% GPU usuage. A 7px grid gives me stutterfree motion @30% CPU usuage. (wow, amazing difference  hmm  )

Cant tell you, why it stutters before the hardware hits its limits (100%) but one can say this: smaller grids are not automatically better. They give little bit more fluidity (theoretically) on the cost of ugly artefacts. The 12px pre-setting is a very good compromise already.


(9 replies, posted in Using SVP)

As long as it takes maximum 90 minutes  to copy a 90 minutes movie to that harddrive, the speed should be fast enough. wink

I imagine one exception. If that HDD is in use (because you download more stuff to that drive at the moment you watch the movie), then it MAY cause problems, depends on the speed you are downloading (More than 100Mbit etc) Maybe you are doing even more like moving stuff from one HDD to another AND downloading AND watching the movie, all from the same device...

In that case I would suggest to copy the movie you are actually watching to a physically different HDD, that is not in use atm. to watch it glitch free, while the other HDD work like hell.

Chainik wrote:

I would like to see a new shader (shader 24 or whatever) that is a 23, but basing on 13 not on 11.

we tried that one long ago and voted to remove it cause it's just ugly  big_smile

hmm. ok

Well Met, grid is not the topic. Settings is not the topic.
The topic is, why does the creator of shader 23 use shader 11 as basis. Why didnt he not use a 13, wich apears superior to 11.
In other words: I would like to see a new shader (shader 24 or whatever) that is a 23, but basing on 13 not on 11.

And I hped you would be gone already. ;-P


(72 replies, posted in Using SVP)

What i meant to say is that i have tested and compared an bluray movie (all2hd+dnm = off / disabled) with an dvd movie (all2hd+dnm= V /enabled ) using windvd ... and i must say i didnt saw much difference quality wise...back then..

Then it was either:

a) you are very far away from the screen
b) you watched something that was converted (SD source that was "professionaly" upscaled and sold with HD label on blue ray) There is lots of this crap on the market. wink

If its a) then, from the pure resolution point of view (there are other "Problems" where these numbers are not enough distance) its said that:

SD material: 4.2x screendiagonale as eye to screen distance = resolution matches resolution limits of a human eye, increasing resolution does not improve quality anymore.
720p: human eye reaches its limits at 2.3x screendiagonale eye to screen distance
1080p: human eye reaches its limits at 1.6x screendiagonale  eye to screen distance
Ultra HD (so called "4K"):  human eye reaches its limits at 0.8x  screendiagonale eye to screen distance
Super High Vision (so called "8K"):  human eye reaches its limits at 0.4x  screendiagonale eye to screen distance

So, how far away are you sitting from your TV? wink


(72 replies, posted in Using SVP)

I just put the 2 images into photoshop and switched the visibility of one layer on and off..... the only thing in these 2 images that changes arevery few pixels at the edge of the dust shields, the rest is not "blinking" (means my eyes cant make out any difference). It apears identical.

Then I mixed them "difference" wise to check the difference. A few (very few) parts of the image differ at about 0,2-0,5%, most parts differ at 0%.

Uh but better stop, this is not the topic here. wink


(72 replies, posted in Using SVP)

SErIOUSLY,  you must have noticed that tooOO .. dont you  hmm

So and again... i PREFER SHARPNESS above anything else ....  Meaning sharper = MORE DETAILS..  big_smile

In that case, you need the least advanced scaling. Because its the sharpest. So more hardware damanding te scaling is, so less sharpness but less aliasing it has. (rough rule)

Oh nose.... SD + Enhancements (i.e desktop resolution + avisynth filtering) to me .. looks as great as HD → whitout enhancements though...

Desktop resolution? Thats SCALING. When ever you scale a video to the desktop resolution, it uses a scaling algorithm. madVR or not, it does use one of the ones madVR offers aswell. So you just need to find the one that windows usual uses. I think DXVA2 is standard (hardware scaling of the GPU) Everything is normaly scaled by that. Games in non-native resolutions and so on. If you think thats the one and only really good working one for you, pick it in madVR....
And f you think SD material with DXVA2 + filtering does look "exactly like real full hd material too you, well, gratulations  lol )

Yes... Croma is more for collor.... while luma brightens the picture...  I have noticed that while using sharpening filter....

You have to be carefull using Chroma.. cuz it tends to create allots of color noise in your pictar playback !!!!!!!

I know about Chroma and Luma, I studied Graphic Design 20 years ago and I am quiet fit with photoshop. Its a usual "trick" to not sharpen Chroma because of noise, while only sharpen Luma. And even there you make a mask that sharpens dark more than light.
The Chroma contains colors without any shading. The Luma shading without any colors.

A Chroma channel looks like that:

While the Luma channel looks exactly like a black+white picture.
Thats a luma channel:


(72 replies, posted in Using SVP)

I dont get what you want to tell me.

Look at those 2 pictures, they are virtually identical, except for those dustshields, wich have slightly lss aliasing (while other parts do have the very same crappy aliasing). I dont see the "Big improvement" that you seem to see.

but in my case for regular dvd .. no matter what renderer ... or setting i use... i dont see any vissual improvement .....

Well, me neither. If I use DivX (or XVid) SD material and scale it up with madVR, no matter what scaling method, they all look like the same identical, total unwatchable crap, wich is only bearable on 16 inch laptop screen in 2m distance but not projected to the wall.

Best solution: Fuck SD, get HD material of everything, if possible. That means: Fuck DVD (SD Quality) wink

Oh yeah and...
Chroma channels are suposedly only 1/4 resolution of Luma (Picture) channel (to get smaller filesizes and chroma is for the colors only while Luma has the details). So they are ALLWAYS scaled. I would say I cant see a difference either if that chroma is scaled bilinear or jinc.


(72 replies, posted in Using SVP)

I mean, i dont see any visuall improvements between madVR and EVR Custom realy...

EVR isnt Windows default renderer, its an improvement already.

As far as I read in other forums, EVR is a huge improvement of image quality compared to windows default renderer while madVR is only a tiny, for most people even not noticable improvement over EVR.

But I agree that upscaling SD material to Full HD, even jinc 8 is just not good enough. Since I use a projector and watch at a screen diagonale of 2 meter, SD material became totaly unwatchable to me. No matter what scaling method used.

btw, this is the difference between Lanczos and Jinc:
Left Jinc, right Lanczos:

As someone once said: the last 10% of quality take 90% of the hardware load.  big_smile


(72 replies, posted in Using SVP)

Just checked that theory myself and changed from madVr to EVR Custom...  and voila ...   NO more frame drops both in window & fullscreen REGARDLESS of which SVP profile settings....

Of course. Its because EVR cant do anything better than bicubic scaling and, even if madVR is set to bicubic scaling aswell, it uses twice the GPU load of EVR. At least for me. (something like 50% GPU load with madVR and 25% GPU load with EVR with identical scaling method chosed)

But EVR has other issues for me (strange doubled tearing test line), thats why I use madVR.


(72 replies, posted in Using SVP)

I can even use smooth motion and get 0 dropped frames"

You dont need smooth motion, because you have something superior: SVP

"Smooth motion" is comparable to "Blend frames to screen refresh rate:true" and both are inferior to real frame inerpolation like SVP makes possible.

SVPs blend to screen refresh for example is more for:
You have a 120Hz monitor but your mashine is too weak for real 120FPS frame interpolation, so you interpolate to 60 and then blend to 120. It makes no sense to real inerpolate and fake blend both to the same value and you dont need fake blend stuff like madVR "smooth motion" if you have real interpolation like SVP.


(72 replies, posted in Using SVP)


(72 replies, posted in Using SVP)


This may sound funny, but .... is SVP started and running in the background?  big_smile
I get a similiar effect (MPC claims that video runs at 120FPS, but full screen does produce thousands of dropped frames in relatively short time (window does not) if I forgot to start SVP and FFDShow forgot to switch of Avisynth without SVP.
So, Avisynth+Reclock running, SVP does not.

And I would not overclock the GPU for madVR, its not worth it (if we assume that OC reduces the lifeexpectancy of the GPU by 50% (if permanently @ OC) (and GPU-developers say GPUs work 4-6 years at default settings before all the heatingup/cooling down destoys them) while most people are unable to see the difference between madVR and EVR-CP. Only between windows default and EVR-CP do most people recognise.