(64 replies, posted in Using SVP)

"type" is not function, it's option setting. Search for SVPAnalyse option
there are several kind of type ( it's in analyse_params ):
1. main.search.type
2. main.search.coarse.type
3. refine.search.type

I would suggest you to modify your script to conform with JSON so it would be easier for you to tweak the setting too.

It's already available for what kind of value for each type (I put here for completion)
2: Hexagon search (most lightweight)
3: Uneven Multi Hexagon (UMH)
4: Exhaustive search (most heavyweight)


(20 replies, posted in Using SVP)

Are you using MadVR? Is this happened when upscaling the video?
If the answer of those two is yes, you need to lower upscale setting of MadVR.

Also, AMD Crimson driver only has trouble playing with some x64 player (using SVPFlow x64). No problem with the usual SVP4. (As long as your driver is not a legacy driver, if it's Crimson beta for legacy driver it's already explained here http://www.svp-team.com/forum/viewtopic … 212#p56212 )


(20 replies, posted in Using SVP)

Did you installed all components from installer package or separate install?

If it's separate install, I would suggest to uninstall all codecs you have and reinstall SVP4 back.


(104 replies, posted in Using SVP)

Your adapter is not legacy adapter smile,
the 7000 series that considered legacy are only up to 7600


(104 replies, posted in Using SVP)

Try to add it to 20/20.


(104 replies, posted in Using SVP)

How many last/ahead buffer did you fill in ffdshow avisynth tab?

Of course you're using Avisynth+ r1779 mt, right?

Nintendo Maniac 64 wrote:

Does this work with 15.11.1?  That's pretty much the go-to version for anyone that's having any issues with Crimson.

Unfortunately I don't have that version. sad
The latest version I have is 15.7.1 (official release) and 15.12 (Crimson beta for legacy adapter)

I guess my problem is in Windows update, because Catalyst never crash before when using Windows 10.
Now, it's always crash.

Of course, SVP and the driver itself is still working flawless smile


(12 replies, posted in Using SVP)

Are you using Avisynth? Is this that you need? Trim

Trim(x, 0, false)

where x is the number of frames that you want to delete.


(7 replies, posted in Using SVP)

hoover1979 wrote:

The resize options in the renderers (EVR (custom presenter), VMR9, MadVR) do not work on my machine and process all videos lower than 1080p pre-resize and then upscaled in shitty Nearest Neighbor.  So I either get Blocky videos or Flickering ones.

Could you give us the screenshot of MadVR statistics whether the upscale working or not. Press Ctrl+J when playing with MPC-HC.

Yeah, I thought of doing it another time big_smile , not at this time thought

Maybe some users here have the same case with me, and they have already installed it. It would be a nice reference if they already did it smile

There's Crimson (forever) beta http://support.amd.com/en-us/download/d … ws+10+-+64

Actually, my AMD catalyst always crash when booting up PC, that's why I'm hesitating between upgrading to that (forever) beta edition or not.
The driver itself is working without any problem.

Thanks for the heads-up.

I have HD56070, and from AMD site it's already classified as legacy driver. The last official release from them for my GPU type driver is 15.7, and AMD provide Crimson edition beta.

My question is, should I upgrade my driver to Crimson Edition? Will I be affected too as the said issues?
If anyone has legacy GPU and has tried the Crimson Edition, please share your experience. Thanks in advance.

hoover1979 wrote:

However I got things working now after first uninstalling MPC-HC, then AVIsynth, then FFDshow, then LAV, then Reclock, then SVP4 Pro and reinstalling SVP4 pro.

Things will work out-of-box if you were installing all necessary components from SVP4 Pro package from the first place.

Conversely the most probable thing you get when installing codecs independently is problem.

That's why codec package available for convenience in order to avoid those kind of problems.


(104 replies, posted in Using SVP)

fuck_yeah wrote:

K-Lite? Does someone still use it?))

I do, for the sake of convenience.

It's just like aptitude or apt-get in Ubuntu Linux or pacman in Arch Linux.
While I still can do without those package managers, it's certainly convenience to have/use those.


(104 replies, posted in Using SVP)

Since you have an i7, the optimum threads is 19.

Have you ever tried to move the plugins64 to folder which not in MPC-BE folder?
In my case, I put the SVP1.dll and SVP2.dll x64 in "C:\Program Files (x86)\SVP\plugins\x64".

What version of MPC-HC do you have? Mine is, see the pic.

p.s.: My MPC-HC (32bit and 64bit) comes from K-Lite and all others necessary codec chain. It works too with MPC-BE (32bit and 64bit) without installation.

Edit: I've tried your script. Except changing width and height from 16 to 32 and disable nnedi3_rpow2, it works without problem both MPC-HC and MPC-BE.


(104 replies, posted in Using SVP)

I've shared my configuration script in this post, http://www.svp-team.com/forum/viewtopic … 563#p55563
My system is far inferior compared to yours, see the pict

Attached too the performance comparison between 32bit vs 64bit, the tested video duration only 01:40 though.

PS: I am using ffdshow from K-Lite codecs.


(104 replies, posted in Using SVP)

fuck_yeah wrote:

But it doesn't work with GPU acceleration when using 64 bit programs (so it's disabled)

What do you mean with that? Do you mean that GPU acceleration is GPU decoding like decoding 8-bit video?
If yes, keep in mind that when using SVPFlow, you already cannot use GPU decoding.

If you mean GPU assisted computation, it works with 64-bit programs/codecs chain. I've been using it for months already.
If it cannot, there's no reason I choose to use SVPFlow x64 over SVP Manager.

While InterFrame seems easy to use based on your script, the moment you use the script it would be meaningless if you have less freedom in tweaking the setting.
I would advise to use SVSuper, SVAnalyse and SVPSmoothFps directly.


(15 replies, posted in Using SVP)

It doesn't crash when I use SVPFlow only. I think the effect is same as "Don't reset" like stated above.

Using Avisynth+ r1779 + ffdshow x64.


(51 replies, posted in Using SVP)

Some peoples here are able to utilize their integrated and discrete GPU.
But unfortunately that's just not anyone can do.

I can't do it, activating discrete GPU will deactivate integrated GPU and vice versa.
I did try activating integrated GPU in bios and not long after, it simply crashed my PC sad

Your problem is the load CPU too high while GPU is not strained at all, is that correct?
I think the key to solve it is in driver, but I'm not sure how to proceed after that since I don't have Nvidia GPU.
How about running DDU once more without reformat your OS?


(11 replies, posted in Using SVP)

maruschade wrote:

No problem using image doubling with source 720x576 even with octuple settings :-)

720x576 upscaled to?

The TS problem is he's upscaling to 2560x1440, not to mention that he's using "Always - if upscaling is needed", so image doubling is always used even though his video already 1080p. (When upscaling 1080p -> 1440p, I don't think he need image doubling though).

How about tweaking the performance tab in madvr?

Nintendo Maniac 64 wrote:

Unless you meant using SVP on 1440p and 2160p video, but so far anime is only ever 1080p (though it is commonly 10bit AVC which can only be decoded in software, so that would result somewhat higher CPU utilization).  For reference HEVC and VP9 shouldn't be much of an issue since the iGPU on Skylake can accelerate the decoding of those video formats.

No, I really meant it. His main videos type to watch is anime, since that kind of video only available 1080p by largest, the more he need to procure more powerful GPU, especially when there's the need to enable Image Doubling (We have the precedent issues about GPU usage, haven't we?).
Let's say if he has 480p videos to be played at resolution 1440p or 2160p. Unless he enables Image Doubling, it can be a bad experience.

And CPU of course, mainly watching anime doesn't necessarily not watching film movies. If there's a chance he has the 4K broadcast, it always nice to get room for performance.

That's why I said, "If larger than that", by "that" I meant was "monitor display resolution 1920x1080"
(I incorrectly posted "size" instead of "resolution" though, maybe that the reason of misunderstanding? smile )

PS: Oh I forgot adding it smile I don't know which CPU/GPU that considered powerful nowadays. So if your chosen CPU/GPU already powerful enough, that's it big_smile If by any means you can procure i7, that would be better

Subjectively speaking, that spec is enough especially if your display monitor size 1980x1080.

If larger than that, I suggest you choose more powerful CPU and GPU.


(11 replies, posted in Using SVP)

Vyphox wrote:

Wouldn't a higher refresh rate mean that more frames will have to be rendered thus increasing GPU usage (with SVP)? Sorry if I'm missing something.

Yes, you're correct. It's not entirely unrelated (read: somehow related), but it's more constraint of CPU. The more larger the frames, the more pixels that need to be computed.
The GPU renderer only works for rendering. Indeed it's increasing GPU usage, but not as much as CPU usage

I understand that upscaling it more will strain the GPU more but I am just confused on how maxing out the settings is less straining than upscaling just to 1440p (can't even handle 16 neurons).

I think it's the Image Doubling is that very straining. Especially in your case which set it "Always - if upscaling is needed".
To be honest, I never be able to use any Image Doubling setting.

If upscaling is actually that much more straining, what would be the better option? 1440p with no image doubling or maximum settings with 1080p? Also, does anyone have an explanation for the dropped frames even though the GPU is only running at 60% usage?

Thank you!

If the video is 1080p, It's better without image doubling. Below than that, it's preferable to use image doubling.
Image doubling gives advantage when upscaling too small videos, e.g. 740x480 -> 2560x1440, of course in exchange with GPU usage

When the frames dropping, is your CPU below 100%? Does that happen continuously?
Sometimes there are scenes that strain CPU more, but it should be only for those kind of scenes. It shouldn't be happened continuously.


(11 replies, posted in Using SVP)

Refresh rate has nothing to do with GPU usage.

Upscaling and downscaling are the issues. The more it upscales or downscales, the more strains are for GPU.

There's nothing abnormal about your issues smile


(2 replies, posted in Using SVP)

If the target frames are exact multiple of video frame rates, the artifacts are certainly reduced.

Setting target frame rates that exceeds monitor refresh rate will result of decreasing the smoothness, it's due dropping frames by renderer.
Not to mention it will burden the CPU usage more while give no advantage over smoothness vs artifacts issues.

(I've experienced it first hand, I though I could setting the target frame rates 3x from 24fps to 72fps, and then using SelectEvery to discard the excess frame rates, the result was disaster lol )

Your best bet to reduce the artifacts are by setting the artifact masking if you're not be able to change your monitor refresh rate (using CRU for example).

I think the "wave-artifacts" is not a bug, it's limitation of implemented algorithm.
This is the computer bug, https://upload.wikimedia.org/wikipedia/ … 96566k.jpg lol