Vizio 2016 D Series Thread

[Voultar]

If you want to be precise you're going to need a 1000fps camera in order to achieve a 1 millisecond resolution per frame of film/video.

OK

I don't understand your cute little eye roll. Is there something that you don't understand about the testing methodology or how it works?

You're wanting to sample latency by generating a tone and calculating the interval (ms) of discrepancy. That's fine. But you're limited to to 1/4 of a frame or 4.16ms of resolution, so that's still an approximation.

I'm only saying that if you want to capture latency via video and wanted 1ms resolution per each frame, you would need to capture @ 1000FPS

[Guspaz]

You can estimate latency with greater accuracy than the interval between frames in the videocamera, such as by using many samples (since each time you record your offset would be different), assuming that you've got a high degree of precision for the actual framerate (in terms of deviation in time between frames). For all practical purposes, however, such precision is not useful for real-world use. If you just throw a 240fps video camera at the problem, is the difference between ~12ms and ~16ms really important? Far more important is the difference between 12ms and 40ms, and a 240fps camera could tell that just fine.

[Voultar]

You can estimate latency with greater accuracy than the interval between frames in the videocamera, such as by using many samples (since each time you record your offset would be different), assuming that you've got a high degree of precision for the actual framerate. For all practical purposes, however, such precision is not useful for real-world use. If you just throw a 240fps video camera at the problem, is the difference between ~12ms and ~16ms really important? Far more important is the difference between 12ms and 40ms, and a 240fps camera could tell that just fine.

I agree. I felt that 8ms resolution per each video frame still was "good enough" for the purposes, here, that's why I shot at 120FPS. The Vizio is positioned lower in frame than the CRT. That too probably adds maybe ~1ms as the fake raster draws the top and then bottom of the picture. But we're really splitting hairs, at that point. lol

[citrus3000psi]

I agree. I felt that 8ms resolution per each video frame still was "good enough" for the purposes, here, that's why I shot at 120FPS. The Vizio is positioned lower in frame than the CRT. That too probably adds maybe ~1ms as the fake raster draws the top and then bottom of the picture. But we're really splitting hairs, at that point. lol

Which is why I recommend putting the CRT next to side of the tv. You might as well get the most accurate number with the equipment you have. I'm not trying to get down to 1ms degrees, but like I said before I was looking for something that gave me more accurate results than 1/2 frame. I wasn't saying your 120FPS is wrong. I've been using a DSLR for most of my testing, and I hate having to guess half frames. Which is why I just bought the mic.

[Voultar]

I agree. I felt that 8ms resolution per each video frame still was "good enough" for the purposes, here, that's why I shot at 120FPS. The Vizio is positioned lower in frame than the CRT. That too probably adds maybe ~1ms as the fake raster draws the top and then bottom of the picture. But we're really splitting hairs, at that point. lol

Which is why I recommend putting the CRT next to side of the tv. You might as well get the most accurate number with the equipment you have. I'm not trying to get down to 1ms degrees, but like I said before I was looking for something that gave me more accurate results than 1/2 frame. I wasn't saying your 120FPS is wrong. I've been using a DSLR for most of my testing, and I hate having to guess half frames. Which is why I just bought the mic.

What model do you have? It's great that people are measuring the different sizes in the Vizio 4K D lineup. I'm even more curious about the D40U-D1, as Bob measured almost 3 frames on that thing. He probably screwed it up. lolz

[Guspaz]

The rolling shutter of the camera can add far more than 1ms. The Sony A7S, for example, has a 24ms rolling shutter delay between the top and bottom of the frame (in full frame HD video), no matter the shutter speed setting. The Canon 5D III isn't much better at 18ms.

[Voultar]

Well the Hi-Def is stripping out color data but I can still get it to sorta work with both composite and HDMI simultaneously, just like Orange808 did.

Hi-Def paired with the D58U. Clear action & game mode enabled via HDMI port #5:

[citrus3000psi]

Well the Hi-Def is stripping out color data but I can still get it to sorta work with both composite and HDMI simultaneously, just like Orange808 did.

Hi-Def paired with the D58U. Clear action & game mode enabled via HDMI port #5:

From that picture it almost looks like 0ms lag. The vizio does have 12-13ms so it must be a rolling shutter like guspaz said. Can you not put the tv to the side, or just a space issue?

I have the D40U-D1. I get about 1 frame lag through an OSSC pushing 720p. I'm using an older Canon DSLR with a 20PVM. Source is fed into the PVM, then the PVM output into the OSSC

Can you do some testing on INPUT 4 with game mode enabled? Curious to know what you get.

[RGB0b]

I'm even more curious about the D40U-D1, as Bob measured almost 3 frames on that thing. He probably screwed it up. lolz

LOL, fuck off, the pics don't lie!!! Now I wish I kept the TV just as proof! My CRT was to the right of the flat-screen when taking pics...not sure if that makes a difference. Also, I used both my iPhone in 240fps, as well as a DSLR on the fastest shutter speed possible. Anything other then input 5 had terrible lag (I think we all agree on that), but I definitely got 2 frames or more on input 5.
Don't forget, I really wanted this TV to be the perfect low-lag solution for me. I was so disappointed when it showed 2 frames that I did the test like 10 different times. Maybe I got a defective set? One with a different panel or board in it?

[orange808]

I've tested a friend's 40" panel and my 50", both have sub-1 frame lag.

This is the first I've heard of two frames or more on input 5. That certainly isn't true on my display. If it were, it would show in tests. Even more damning, it would be immediately apparent when I jumped off the coinbox to the flag in world 4. If I don't get 5,000 points consistently, something is wrong.

If there is a revision of the panel that is significantly slower, that would be interesting and valuable information. It seems more likely that retrorgb got a bad unit.

[paulb_nl]

From that picture it almost looks like 0ms lag. The vizio does have 12-13ms so it must be a rolling shutter like guspaz said.

13ms lag is measured from the middle of the screen and is including the display draw time. Since it takes around 8ms to draw from top to the middle you have to deduct that from the leobodnar result to get the true input lag. So the true input lag of the Vizio compared to the crt should be around 5ms.

[bobrocks95]

Looks like this would be a good stopgap between 1080p sets and 4K OLED sets, while we wait for lower lag and prices on those.

I'm interested in HDR as well, but afaik that has lag problems of its own. Freesync 2 may fix that in the PC space, but I wonder how long it will be before TVs have HDR modes that don't crank the lag up? Unless I'm blowing it out of proportion and all the HDR TVs available now are just laggy to begin with?

Glad that there's a reasonably priced option for low-lag right now.

[Voultar]

Well the Hi-Def is stripping out color data but I can still get it to sorta work with both composite and HDMI simultaneously, just like Orange808 did.

Hi-Def paired with the D58U. Clear action & game mode enabled via HDMI port #5:

From that picture it almost looks like 0ms lag. The vizio does have 12-13ms so it must be a rolling shutter like guspaz said. Can you not put the tv to the side, or just a space issue?

I have the D40U-D1. I get about 1 frame lag through an OSSC pushing 720p. I'm using an older Canon DSLR with a 20PVM. Source is fed into the PVM, then the PVM output into the OSSC

Can you do some testing on INPUT 4 with game mode enabled? Curious to know what you get.

It's really a space issue. This is a 36" Trinitron thar I RGB modded years ago, hard to move a TV that's nearly 60 inches next to it.


But like I said before, the Leo Bodnar relies on pixel fill rate and panel response in its testing methodology, there are no intermediaries. So coupling that with the fact that each frame of my video is an 8.3ms sample, this appears to be quite accurate. I can upload the video if anyone wants to see it.

[mikejmoffitt]

We all know the light gun systems generally don't work on an HDTV for timing issues. If the light sensor is able to work at all, though, to detect a white screen versus a black one, it would not be impossible to make an NES program that uses two light guns and turns the screen white, then measures the amount of time between one Zapper detecting it and the other Zapper. Using two, pointed at the same spot on the screen (assisted by the program), a direct lag test could be made to compare two displays, which would help work around the various problems with using cameras (obtaining high speed devices, dealing with rolling shutter, and the beat frequency from the camera's capture rate and the display's output rate not being genlocked). Something like that would allow for a definitive answer.

[Voultar]

We all know the light gun systems generally don't work on an HDTV for timing issues. If the light sensor is able to work at all, though, to detect a white screen versus a black one, it would not be impossible to make an NES program that uses two light guns and turns the screen white, then measures the amount of time between one Zapper detecting it and the other Zapper. Using two, pointed at the same spot on the screen (assisted by the program), a direct lag test could be made to compare two displays, which would help work around the various problems with using cameras (obtaining high speed devices, dealing with rolling shutter, and the beat frequency from the camera's capture rate and the display's output rate not being genlocked). Something like that would allow for a definitive answer.

The Zapper is basically just a photo transistor, so that would probably work okay. You could make a cartridge with timers in there and the NES code would know where the refresh is. So by counting cycles before the transistor registers you could theoretically calculate how many ms of lag there is. Just count from VBLANK to a particular point. You wouldn't even need to know where the white square is outputting, you know where this is because it's X scanlines after VBLANK, so it becomes a fixed offset. You could test 3 points on the screen; Top middle and bottom. And knowing the offset you could subtract this providing that you started from VBLANK.

[Guspaz]

You shouldn't need two zappers, or even a CRT: the zapper can be polled each scanline, the NES knows what scanline it's drawing when the zapper sees the light, and there shouldn't be any real delay between the NES drawing a scanline and the scanline appearing on a CRT.

Put the zapper at the top of the screen, pull the zapper trigger, NES changes the background colour from black to white, NES times how long it takes for the zapper to detect the light. There, you've got your latency to within a few scanlines.

EDIT: Admittedly I'm ignorant of the NES architecture. There are multiple ways to do this on the Gameboy, including at the minimum reading the scanline register, setting an hblank interrupt, or setting an LYC interrupt, so I'm assuming that there are similar ways to do this on the NES.

EDIT2: Looks like the NES doesn't have any of those things, but MMC3 has a scanline counter that can be used for that, and most flashcarts should be able to provide that.

[mikejmoffitt]

You shouldn't need two zappers, or even a CRT: the zapper can be polled each scanline, the NES knows what scanline it's drawing when the zapper sees the light, and there shouldn't be any real delay between the NES drawing a scanline and the scanline appearing on a CRT.

Put the zapper at the top of the screen, pull the zapper trigger, NES changes the background colour from black to white, NES times how long it takes for the zapper to detect the light. There, you've got your latency to within a few scanlines.

EDIT: Admittedly I'm ignorant of the NES architecture. There are multiple ways to do this on the Gameboy, including at the minimum reading the scanline register, setting an hblank interrupt, or setting an LYC interrupt, so I'm assuming that there are similar ways to do this on the NES.

EDIT2: Looks like the NES doesn't have any of those things, but MMC3 has a scanline counter that can be used for that, and most flashcarts should be able to provide that.

Both methods could be implemented. However, the intent is to convey the difference between a control monitor and the tested monitor, as a regular CRT will have a few milliseconds, varying further by how far down the screen and to the side the Zapper is pointed, so having the two-Zappers method would be important to me as well.

I want to do this mostly because I have an HD CRT (KV-32HS510) and when fed with an XRGB3 in line doubling mode, or 480p straight out of one of my FPGA-equipped consoles, the latency is very low but I'm having trouble proving it with a 240fps camera. Sometimes it looks like it's on time, sometimes it looks delayed by a frame. So, I want to put together something reliable and empirically reproducible, and Zappers aren't too expensive.

[Guspaz]

The variance with the Zapper pressed up against the screen would only differ by how much overscan the display has, which should be a relatively minor difference, and while there are likely many people out there who have a zapper and some way of getting ROMs onto an NES or NT Mini, there are probably much fewer people who have multiple zappers.

[mikejmoffitt]

The variance with the Zapper pressed up against the screen would only differ by how much overscan the display has, which should be a relatively minor difference, and while there are likely many people out there who have a zapper and some way of getting ROMs onto an NES or NT Mini, there are probably much fewer people who have multiple zappers.

Pointing at the top of the screen is not a good way to do it. Why would you intentionally allow variance from overscan when you can have the user center on a target at a known location first?

Again, a program like this could support two methods. I am only interested in the direct comparison using two, but the cost of implementing both is negligible, so it might as well be done.

Many people who play Super Smash Brothers competitively will say some ridiculous things - "This CRT lags more than that one" "everyone knows all CRTs have different amount of lag" "my sick LCD setup has less lag than a CRT!" - and rather than try to explain how monitors work and the impact of an unbuffered signal path, it would be great to just pull out empirical evidence so they can stop whining and just play on the monitor without blaming their loss on "the lag".

Anyway, I don't mean to de-rail the thread which was discussing a specific TV series, so if I or any one actually makes this test I'd throw it in its own thread.

[orange808]

It could still be overclockable and forced to display in 120Hz.

Been working on my PC game backlog. Finally got around to hooking up my game rig and forcing 120Hz on the Vizio. Yes, the Vizio can display an image with 120Hz sources, but it's not good.

At first, I thought thr problem might be: using ordinary vsync (instead of gsync on my monitor), but it's more than that.

Blur Busters has a very nice test pattern for overclocked screens and the Vizio failed miserably.

At 120Hz, the display skips frames. It can accept 120Hz sources and put an image on the screen, but it drops frames.

Doesn't work properly.

It generally skips every other frame, so there is very little obvious motion judder; that's probably how these Vizios are fooling people into believing they can do 120Hz properly.

[Xyga]

Wasn't it only the P and M actually supporting 1080p@120 ? (and not all sizes iirc)

[Bahn Yuki]

It could still be overclockable and forced to display in 120Hz.

Been working on my PC game backlog. Finally got around to hooking up my game rig and forcing 120Hz on the Vizio. Yes, the Vizio can display an image with 120Hz sources, but it's not good.

At first, I thought thr problem might be: using ordinary vsync (instead of gsync on my monitor), but it's more than that.

Blur Busters has a very nice test pattern for overclocked screens and the Vizio failed miserably.

At 120Hz, the display skips frames. It can accept 120Hz sources and put an image on the screen, but it drops frames.

Doesn't work properly.

It generally skips every other frame, so there is very little obvious motion judder; that's probably how these Vizios are fooling people into believing they can do 120Hz properly.

Fooling who? The D series are 60hz panels. You need M or P for 120hz.

Sent from my SAMSUNG-SM-G870A using Tapatalk

[lechu]

Semi off topic, but has anyone had experience with the 2017 D series? I'm having an easier time finding this year's models rather than last year's.

[Blair]

Semi off topic, but has anyone had experience with the 2017 D series? I'm having an easier time finding this year's models rather than last year's.

easier in what way?

[lechu]

Semi off topic, but has anyone had experience with the 2017 D series? I'm having an easier time finding this year's models rather than last year's.

easier in what way?

I guess I should rephrase that. Easier for my needs and cost effective I guess.

The VIZIO D43-E2 43" 4K (2017) is just shy of $600 CAD. VIZIO D40u-D1 40'' 4K (2016) is almost $900 CAD. Hell, the VIZIO D40-D1 40'' 1080p (2016) is the same price as the 2017 4K TV and is ever so slightly smaller. If they're essentially the same TV or the difference is marginal, I'd rather save $300.

[Ashura]

What is the reasoning behind port 5 being the magical HDMI port on these displays? Reading the manual, it's listed as the only 4K 60hz port. I'm a bit dumb about 4K tvs; why so many differently rated HDMI ports? Do the other ports not even do 60hz on 1080p or less content? Is it just a cost thing?