4K Blues

[Woozle]

https://imgur.com/a/y859w hi def nes on a Vizio 4K set. My Dad lives 4 hours away otherwise I could take more pics. I think it looks great, I was surprised by the lack of blur/ghosting.

One thing that would be great with 4K TVs is that 240p content can be scaled 10X, giving room for perfect scanlines and razor sharp sprites. With 1080p, you'll either have 4X scaling which doesn't fill the screen or 5X scaling which gives an uneven amount of scanlines. I'd buy an XRGB-4K anyday.

I think scanlines at 5X look amazing with 3 on/2 off. I prefer it to scanlines at 2X, 3x or 4x. https://imgur.com/a/qfi5H That's the GBA with 5x scale in a 720p window.

[ZellSF]

I don't suppose anyone with the HDFury Linker, iScan Micro or Mini has a 4K capture card? A sample image could be useful for 4K owners to determine if a scaler would be an upgrade or not.

[orange808]

Thanks for the pics! The Linker results look good. Does it keep a RGB 4:4:4 output when the input is RGB 4:4:4 ?

If the display supports it, yes. It's a 4k and 1080p swiss army knife.

I don't suppose anyone with the HDFury Linker, iScan Micro or Mini has a 4K capture card? A sample image could be useful for 4K owners to determine if a scaler would be an upgrade or not.

I have no capture card. Can't help you there.

------

Also:

The Linker does a nice job scaling. No kidding. With that said, it only processes 1080p and 4k input signals; everything else simply passes through. That means that 4k panel owners that buy a Linker to bypass a rubbish OEM scaler will probably need another scaler/video processor in the chain.

console -> 240p -> line doubler -> 480p -> scaler -> 1080p -> Linker -> 4k

With the OSSC, an iScan vp model, and the Linker, it's probably ~8ms of lag.

[Fudoh]

The Linker does a nice job scaling. No kidding.

how do you actual video sources look like ? Your screenshots above look great, but it's a low res game. I had the HDFIV a few years ago and wasn't thrilled by its scaling.

With the OSSC, an iScan vp model, and the Linker, it's probably ~8ms of lag.

Is it confirmed that the Linker is basically lag free ?

[orange808]

The Linker does a nice job scaling. No kidding.

how do you actual video sources look like ? Your screenshots above look great, but it's a low res game. I had the HDFIV a few years ago and wasn't thrilled by its scaling.

With the OSSC, an iScan vp model, and the Linker, it's probably ~8ms of lag.

Is it confirmed that the Linker is basically lag free ?

The Linker has been living on my Sony 1080p to clear up some HDCP nonsense. I never planned on watching any video on the Vizio--or using the upscaling on the Linker. My video sources are streaming and an HDHomeRun, so I don't have any pristine 1080p video inputs to showcase. Any suggestions? I could chain something through the DVDO, but that adds another variable. I can try some DVDs through the DVDO; I have a bluray, but I never got around to buying any bluray disks.

I need an Integral to get the Leo Bodnar to work with the Linker. I haven't gotten an Integral, but it looks like I need one so I can keep using the Sony to watch video. I did try the Dreamcast 240p lag test (via microphone) from the DVDOvp50 and I didn't detect any additional lag--versus Dreamcast directly connected to the vp50. 2ms for the Linker is just a guess until I get an Integral.

edit: Got an idea. I can hook up the PC and get some shots of a game upscaled from 1080p.

[Fudoh]

Another question on the Linker: can you check if it provides a framelocked output or if it does a framerate conversion to a standard rate ? If you have a DVDO you can easily check this by feeding the test pattern with the scrolling bar and adjusting the output framerate slightly between 59.8 and 60.2 Hz. At this range your TV on its own should be able to display a smooth scrolling. If the Linker is framelocked it would be the same in 4K. If not, it will introduce an occasional stutter every second or so.

[orange808]

Another question on the Linker: can you check if it provides a framelocked output or if it does a framerate conversion to a standard rate ? If you have a DVDO you can easily check this by feeding the test pattern with the scrolling bar and adjusting the output framerate slightly between 59.8 and 60.2 Hz. At this range your TV on its own should be able to display a smooth scrolling. If the Linker is framelocked it would be the same in 4K. If not, it will introduce an occasional stutter every second or so.

Ok. I went to test patterns and selected "Judder". A white vertical bar sweeps the screen horizontally.

As I incremented and decremented the framerate, the screen would go black for 3 seconds before the test pattern reappeared.

The vertical bar continued to sweep smoothly with no hitching or studdering--after the image reappeared each time.

[Fudoh]

sounds good - thanks for testing!

[orange808]

Finally got the Leo Bodnar to chain through the Linker and measure the lag when scaling 1080p to 4k.

Scaling to 4k with the Linker and plugging the Leo Bodnar in directly (1080p) return identical results.

I can't measure any lag within the Bodnar's accuracy (0.1ms).

[BONKERS]

This is what happens with scaling at such high ratios with low res 2D Sprite based or low res 3D rendered games.

The higher the ratio, the worse the upscaling looks.

That's not how it works.

You have to redraw a image, the bigger the surface you get to redraw it to, the more accurate you can make the redrawing.

That a 4K scaler looks worse than a 1080p scaler is not "what happens", it's a serious defect in the scaler.

1080p alone is 2x2 and that looks bad on most 4k TV sets with 1080p console games. (Play a PC game on a 1080p set with a custom 960x540 resolution without AA or with just FXAA and you'll get a similar idea of how it looks)

No you will not. A 960x540 image scaled to 1920x1080 looks nowhere similar to a 1920x1080 image scaled to 3840x2160.

There's a 4X difference in both the quality of the source image and the size of the surface you redraw it to. Both very important factors for scaling quality.

Do you think a DVDO processor that outputs 1080p would help with bad 4k scalers? In theory, converting 1080p to 4k should be (relatively) trivial. Of course, I'm betting the television will not apply some kind of unnecessary filter to all upscaled video--regardless of the situation--so I might (just) be naive...

TVs probably upscale using their upscaling algorithm no matter the circumstances, they won't intelligently pick a integer scale factor (and one isn't always desirable) no.

You'll just end up introducing one more lossy (and laggy) scaling step. The end result can still potentially look better, but I wouldn't say it's a good idea.

Now a XRGB-mini or OSSC that can do integer scaling that's better for 2D scaling might end up with a better end result, since they don't actually distort the image as part of their scaling process. I don't think it looks good for 3D content though, but might be preferable to your TV's scaling.

UH YES, it is. Do you not realize that 960x540 to 1920x1080 is EXACTLY the same ratio as 1920x1080 is to 3840x2160?

960*2=1920
540*2=1080

1920*2=3840
1080*2=2160

They are both the same 4x increase in pixels.
Thus, 540p to 1080p with ugly scaling will give you an idea of how bad 1080p could look on some 4k TVs. (Worse for lower res content)
The scaling factor is exactly the same. And my comparison is completely relevant. Because
A.) Most 1080p 3D rendered console games do NOT have great image quality. They use basic Post Processing AA, none at all or poor temporal AA solutions (Bar a few exceptions). With very poor texture filtering.

(We aren't talking 2D games, because the majority of 1080p rendered games are 3D in real time.)

This is exactly why the PS4Pro exists, to give developers more options to get better upsampling and uprendering. To get as close as possible to native 4k. Because upscaling from 1080p, with the image quality of the average game 1080p game to 4x. Looks like shit. Which is exactly what they showcased when showing off the PS4P. (You might see some developers targeting a base resolution as high as say 3200x1800. Which gets you to start with before attempting to interpolate between the gaps with CBR or a custom solution. The same scaling ratio as 1600x900 is to 1920x1080. 1.4x1.4.)

http://screenshotcomparison.com/comparison/189038
Here's a comparison using, what is probably the best looking pure 2D upsampling (That isn't point scaling) at 2x2(4x) scaling with and without good AA in the base image. The image with AA actually doesn't look too bad, nice even with 4x scaling. But this is impractical if not impossible to run in real time. And 0% of console games have image quality at their native resolution even remotely this good. So you'd end up with a result more like the first image even if it was possible. (3D games are a whole different can of worms when it comes to upscaling)

And then, even with AA here's what it would look like upscaled if you changed the factor to 16x resolution instead (4x4. Equivalent if you were doing 960x540 to 3840x2160 with RT3D content) and you end up with what some 4k TVs will look like with sub 720p content. A smudged mess. (Though probably not as bad as this looks. Slightly better)
http://u.cubeupload.com/MrBonk/04002C18xgUpPhotosca.jpg
The same 16x with linear sampling
Clearly it's better than linear sampling, but that's not saying too much.

Source Image http://u.cubeupload.com/MrBonk/ef104002C18xg.png

I dont' know how you can sit there and say somehow the higher the scaling factor the more accurate it can look to the original image.
I've never seen a single method that can do it without bringing emulating an actual low res display into the mix(more than just scanlines). And no TV manufacturer would ever bother to implement such a thing for scaling low res content on 4k TVs

It simply does not exist for real time 3D content. That's for sure.

You have to redraw a image, the bigger the surface you get to redraw it to, the more accurate you can make the redrawing.

The more accurate you can make the re drawing? Good luck ever getting any TV that actually achieves that. (Especially with real time rendered 3D content example as above)
In reality, with 99% of current scaling technology, unless you are doing integer point scaling, or emulating the actual underlying display of a lower res display like a CRT or low res LCD (Which actually does benefit with larger scaling). The higher the ratio, the worse it looks. And doesn't ever come remotely close to looking as good as it does on a native res display. OR on a display that has a lower scaling ratio when using the same scaling algorithms.


Having the simple option for point scaling with inputs that are integer multiples of 4k, would make a huge difference for 2D games. Low res and aliased 3D is going to look pretty bad no matter what. But TV manufacturers and even PC GPU vendors don't get it. It's the simplest scaling you could possibly do. And will make 2D games look great.


2D stuff really doesn't need much more than point integer scaling and CRT emulation. But again, that's not all that exists out there.

[ZellSF]

UH YES, it is. Do you not realize that 960x540 to 1920x1080 is EXACTLY the same ratio as 1920x1080 is to 3840x2160?

Same scaling ratio is pretty irrelevant, you're still comparing a low quality image (960x540) to a high quality one (1920x1080). You are not capable of filtering out all the non-scaling differences in image quality to only focus on the scaling factors.

http://screenshotcomparison.com/comparison/189038
Here's a comparison using, what is probably the best looking pure 2D upsampling (That isn't point scaling) at 2x2(4x) scaling with and without good AA in the base image. The image with AA actually doesn't look too bad, nice even with 4x scaling. But this is impractical if not impossible to run in real time. And 0% of console games have image quality at their native resolution even remotely this good. So you'd end up with a result more like the first image even if it was possible. (3D games are a whole different can of worms when it comes to upscaling)

And then, even with AA here's what it would look like upscaled if you changed the factor to 16x resolution instead (4x4. Equivalent if you were doing 960x540 to 3840x2160) and you end up with what most 4k TVs will look like with sub 720p content. A smudged mess. (Though probably not as bad as this looks. Slightly better)
http://u.cubeupload.com/MrBonk/04002C18xgUpPhotosca.jpg
The same 16x with linear sampling
Clearly it's better than linear sampling, but that's not saying too much.

Source Image http://u.cubeupload.com/MrBonk/ef104002C18xg.png

I dont' know how you can sit there and say somehow the higher the scaling factor the more accurate it can look to the original image.

You're looking at the 16x factor resolution at a different ppi than the 4x factor. If you're not, then you're doing another scaling step. Or you're looking at another image entirely. To accurately compare something, you can't throw these kinds of variables in.

You need two different resolution, but equal sized monitors. Get a image that's subnative to both of them and scale them using the same algorithm (using a PC GPU would be ideal).

[ZellSF]

But if you're going to try suboptimal comparisons, I figure I should too. Here I have two images. The source image for both is 480x270 (and in fact part of your source image). One image is bilinearly resized to 1920x1080. The other is bilinearly resized to 960x540, then pixel doubled. Since the pixel doubling is lossless, the only difference here is the amount of room the bilinear scaler has:

http://i.cubeupload.com/ikOBYS.png
http://i.cubeupload.com/W0gLAN.png

Yes, the image that's only bilinearly resized to 960x540 is definitely sharper. Better looking though? More accurate? I don't think so. A smudged mess like you claim? Hell no. And I'm sitting two feet away from a relatively low ppi screen...

Again this is also a very flawed comparison, just better than yours.

Same concept with some pixel art:
http://i.cubeupload.com/TJDkTu.png
http://i.cubeupload.com/t2h5oW.png
Sharper is definitely not better here.

[orange808]

Well, when you two get finished, the problem with OEM scalers is the lack of options--not the display or source.

[orange808]

Spotted a Kanex RGBHV (VGA) to HDMI transcoder/4k scaler I didn't know existed. It's fairly cheap and I have been extremely disappointed with the VGA to HDMI transcoders (no scaling) I have tried (save the DVDO machines).

http://www.kanexpro.com/item/?id=CON-VGA-HD4K

I'll hook up the XRGB-2 to it and try to share some details when I get it.

Looks like they have a component version as well, but I don't plan on getting it.

[ZellSF]

Well, when you two get finished, the problem with OEM scalers is the lack of options--not the display or source.

Repeating myself here, but I really doubt TV manufacturers want more confusing options for their users. You're better off asking for better scaling algorithms.

[Bahn Yuki]

Well, for reference, here's what 720p Hi Def NES looks like on a 2016 Vizio D50u 4k. All I have is a cell phone, but I think camera conveys the picture reasonably.

Scanlines tend to mask certain issues with upscaling, just see the SLG-in-a-Box/GBS-8220! According to [urlhttp://www.rtings.com/tv/tests/inputs/input-lag]rtings[/url], none of the 2016 Vizio models support RGB 4:4:4 (all HDMI inputs chroma subsample to 4:2:2). What does your set look like with scanlines disabled (camera focused) playing SMB1 with the default palette?

That's actually being rectified in a firmware update. For some reason the TV's VPU is taking 4:4:4 and turning it into 4:2:2. in the next update they'll allow the 4:4:4 to bypass the VPU altogether.

No I don't work for Vizio, I just like their displays for gaming. Oh and Dolby Vision, that cannot be understated.

[orange808]

Spotted a Kanex RGBHV (VGA) to HDMI transcoder/4k scaler I didn't know existed. It's fairly cheap and I have been extremely disappointed with the VGA to HDMI transcoders (no scaling) I have tried (save the DVDO machines).

http://www.kanexpro.com/item/?id=CON-VGA-HD4K

I'll hook up the XRGB-2 to it and try to share some details when I get it.

Looks like they have a component version as well, but I don't plan on getting it.

No huge surprise; we all hate no name Chinese video gear.

I gave the Kanexpro VGA to 4k scaler/transcoder a try.

The lag is ~85ms (5 frames). No matter what I did, colors always appeared washed out.

The scaling cannot properly handle scanlines; lines appear uneven.

Absolutely awful. I won't be uploading photos; it's packed up and going back in the morning.

[Shuco13]

Anybody knows a good review of the HDfury Linker? I think of buying one. Thanks.

[orange808]

Anybody knows a good review of the HDfury Linker? I think of buying one. Thanks.

How are you planning on using it?

You might be better off buying a cheap DVDO Micro and waiting for a better solution. It depends.

What devices are in your video chain? What's your budget? What resolutions are giving you issues?

[Shuco13]

I will mainly need it for downscaling but I'm also interested in it's upscaling capabilities.

[LDigital]

Do we have any solid lag/input delay numbers on the DVDO micro? I haven't been able to find any

[orange808]

Do we have any solid lag/input delay numbers on the DVDO micro? I haven't been able to find any

A microsecond or two.

[Blair]

Do we have any solid lag/input delay numbers on the DVDO micro? I haven't been able to find any

I believe with all of the processing options turned off, it should scale progressive sources to 1080p at 6ms, unfortunately I don't have any idea if that holds true when scaling things to 4k. (it outputs 4k60hz 4:2:0, as its highest resolution).

here are the image processing options, as recommended by DVDO.

Off – For video games & animation
Low – For medium quality video (soft 480p movie sources)
High – For low resolution internet & low quality video

(Some believe the "high" setting adds an additional 6ms of lag, but I haven't confirmed that).

[orange808]

Do we have any solid lag/input delay numbers on the DVDO micro? I haven't been able to find any

I believe with all of the processing options turned off, it should scale progressive sources to 1080p at 6ms, unfortunately I don't have any idea if that holds true when scaling things to 4k. (it outputs 4k60hz 4:2:0, as its highest resolution).

here are the image processing options, as recommended by DVDO.

Off – For video games & animation
Low – For medium quality video (soft 480p movie sources)
High – For low resolution internet & low quality video

(Some believe the "high" setting adds an additional 6ms of lag, but I haven't confirmed that).

Chaining the Leo Bodnar devices through the Micro showed a microsecond of lag with 720p and 1080p signals. 480p might be an extra microsecond--and it might not.

The useless "enhancement" feature did not affect lag at all.

Regardless, it isn't 6ms.

[Shuco13]

So no statement on the linker?

[orange808]

So no statement on the linker?

I like the Linker and I use it to downscale in my convoluted video setup, but that falls outside the scope of this board.

When I tried the upscaling, I was impressed. It reproduced the 1080p image faithfully without a "vasoline" interpolated look. I couldn't detect any lag, it didn't create ringing, and the Linker doesn't claim to (or try to) "enhance" the image.

Plugging into a computer lets you select and force chroma settings--if you are inclined to do so.

With less than a microsecond of lag and clean downscaling/upscaling, I think it could (potentially) extend the life of a 1080p video processor or display.

I don't know of any reviews, but HDFury does pay attention to the avsforums.

[Blair]

Chaining the Leo Bodnar devices through the Micro showed a microsecond of lag with 720p and 1080p signals. 480p might be an extra microsecond--and it might not.

The useless "enhancement" feature did not affect lag at all.

Regardless, it isn't 6ms.

that's very strange, almost everything in the DVDO line of processors since at least the HD and HD+ models have processed 480p, 720, and 1080p signals with at least 6 milliseconds of input lag. somehow they managed to get the processing down to a microsecond? that's practically nothing!

how can that be possible?

(unless I'm wrong about the measurement, I always thought 1 millisecond was equal to 1000 microseconds)

I like the Linker and I use it to downscale in my convoluted video setup, but that falls outside the scope of this board.

actually downscaling has been discussed quite a few times on this board, one of the first discussions I ever had on this board was about using video processors to downscale video signals to 240p for use on PVM monitors and consumer CRT televisions. Fudoh has also written a a few posts on the subject.

so you are using the linker to downscale 4k to a 1080p display? from a PC?

[orange808]

Chaining the Leo Bodnar devices through the Micro showed a microsecond of lag with 720p and 1080p signals. 480p might be an extra microsecond--and it might not.

The useless "enhancement" feature did not affect lag at all.

Regardless, it isn't 6ms.

that's very strange, almost everything in the DVDO line of processors since at least the HD and HD+ models have processed 480p, 720, and 1080p signals with at least 6 milliseconds of input lag. somehow they managed to get the processing down to a microsecond? that's practically nothing!

how can that be possible?

(unless I'm wrong about the measurement, I always thought 1 millisecond was equal to 1000 microseconds)

I like the Linker and I use it to downscale in my convoluted video setup, but that falls outside the scope of this board.

actually downscaling has been discussed quite a few times on this board, one of the first discussions I ever had on this board was about using video processors to downscale video signals to 240p for use on PVM monitors and consumer CRT televisions. Fudoh has also written a a few posts on the subject.

so you are using the linker to downscale 4k to a 1080p display? from a PC?

Yes. The lag time of the DVDO Micro is almost impossible to detect. The same can be said for the Linker. Keep in mind that neither unit is a video processor or transcoder. They just scale progressive signals.

Is it amazing? No. My 2016 Vizio 4k D series also scales 720p and 1080p to 4k almost instantly. I can't test the 480p directly, but I bet it is just as fast.

The DVDO Micro is plenty fast, but the ringing issue is difficult to explain.

As for downscaling 4k to 1080p, I can't see how that applies to gaming. Real 4k gaming doesn't exist. (Using an upscaler isn't 4k, Sony.)

[brentsg]

Hey team display guru..

So I've migrated my display from a Kuro to a new LG 55E6P. I really, really like it for most things. Here are my 2 (predictable) issues:

1) The scaler sucks. I have a Framemeister but the display will still need to get things to 4k. I am considering a Linker for this, and I see it's somewhat reasonably discussed.
2) The new HDR game mode is great, but normal game mode is locked to wide color gamut. It looks terrible as expected. Maybe someday LG fixes this but I'm looking for things I can do. I am wondering if the HDFury Integral will be a tool that might help with this, either by mapping colors in a more acceptable way.. or by tricking the display into HDR Game Mode so I can manipulate the display's color gamut.

Thoughts? I'm honestly really stretching my knowledge here and I'm not completely comfortable with the capabilities of the Integral and Linker.

As an aside, I'll likely be picking up the new Oppo 203 when it's released. If it's anything like the 103 then I'll be running my DirecTV thru the Oppo as a poor man's video processor.

Note: I made a thread at AVS as well, in the E6 thread. I expect to get more here, but thought I'd mention it.

[orange808]

Hey team display guru..

So I've migrated my display from a Kuro to a new LG 55E6P. I really, really like it for most things. Here are my 2 (predictable) issues:

1) The scaler sucks. I have a Framemeister but the display will still need to get things to 4k. I am considering a Linker for this, and I see it's somewhat reasonably discussed.
2) The new HDR game mode is great, but normal game mode is locked to wide color gamut. It looks terrible as expected. Maybe someday LG fixes this but I'm looking for things I can do. I am wondering if the HDFury Integral will be a tool that might help with this, either by mapping colors in a more acceptable way.. or by tricking the display into HDR Game Mode so I can manipulate the display's color gamut.

Thoughts? I'm honestly really stretching my knowledge here and I'm not completely comfortable with the capabilities of the Integral and Linker.

As an aside, I'll likely be picking up the new Oppo 203 when it's released. If it's anything like the 103 then I'll be running my DirecTV thru the Oppo as a poor man's video processor.

Note: I made a thread at AVS as well, in the E6 thread. I expect to get more here, but thought I'd mention it.

Well, first of all, I'm going to be jealous of your Oppo.

Second, how granular are the settings on your tv? The HDFury can't force the display to cooperate. What HDR sources are you hooking up?

The Integral allows you to freely change HDCP versions, change chroma/bit depth, and it works as a splitter.

The Linker allows you to change chroma/bit depth, updates older HDCP to 2.2, and upscales/downscales between 4k and 1080p.