Seiki SE50UY04 3840x2160 50" TV ($1300)

I'm sad to inform that the 39 inch model's mini lvds interface is only 2x 81 pins instead of 4 x 81 pins.
The 50 inch panels however will be pin to pin compatible.
I'll strip down the monitor gurther tonight hoping that the other 2x 81 pins are just hidden.
So with a longer ribbon that you can position it will work woth the 120Hz prototype Tcon.
 
I'm sad to inform that the 39 inch model's mini lvds interface is only 2x 81 pins instead of 4 x 81 pins.
The 50 inch panels however will be pin to pin compatible.
I'll strip down the monitor gurther tonight hoping that the other 2x 81 pins are just hidden.
So with a longer ribbon that you can position it will work woth the 120Hz prototype Tcon.

Will these be up for sale at some point? I'm still on the fence about buying one of the seiki's (most likely the 39" but I may swing a few more hundred and go for the 50" to replace my 5x1 Portrait 23" 1080p setup) I'd at least like 60Hz, but 120Hz would be stellar.
 
Will these be up for sale at some point? I'm still on the fence about buying one of the seiki's (most likely the 39" but I may swing a few more hundred and go for the 50" to replace my 5x1 Portrait 23" 1080p setup) I'd at least like 60Hz, but 120Hz would be stellar.
120Hz is doable on the 50 inch depending on if it has 2 hidden unused ports the 39 inch might to.
I figured the 39 and 50 inch used the same Tcon well it turns out they do in fact use the same chip but the 50 inch has 4x 81 to pnael where the 39 only has half that.
Which is strange considering they have the same amount of pixels and the same refreshable rate.

If I get them working I'll of course make them available.
 
Maarten, what do you mean by that? you checked out the tcon and there were 2x 81 pins?
 
Last edited:
I apologize if this question has been answered already, but is the 50" version OK with 1920x1080 @ 120hz (i.e. no frame skipping)? I see issues with the 39", but I'm not sure if they also apply to the 50"
 
I apologize if this question has been answered already, but is the 50" version OK with 1920x1080 @ 120hz (i.e. no frame skipping)? I see issues with the 39", but I'm not sure if they also apply to the 50"

The 50 inch doesn't have the problem
 
I apologize if this question has been answered already, but is the 50" version OK with 1920x1080 @ 120hz (i.e. no frame skipping)? I see issues with the 39", but I'm not sure if they also apply to the 50"

The 50 inch will frame skip at > 120 Hz but its good up to 120Hz. I run stepmania and quakelive at 120Hz and it is smooth as silk with very low input lag. Nothing like the 39 inch version =(.
 
So just to clarify where things stand, at the moment with current firmware:

50" can do 30Hz at 4k, or 120Hz at 1080p

39" can do 30Hz at 4k, or 60Hz at 1080p

Is that accurate?
 
Just to add the 39 inch will accept a 120Hz signal just fine (and list 120Hz on its OSD) so the interface can handle it just fine but the panel is only refreshing at 60Hz. Seiki has promised a firmware update for the 39 inch to fix this as they basically are falsely advertising the 39 inch model as 120Hz right now (as it doesnt upconvert 4k to 120hz at all, and neither does the 50 inch).

I am getting annoyed at how long I have had to wait for this firmware update...
 
Just to add the 39 inch will accept a 120Hz signal just fine (and list 120Hz on its OSD) so the interface can handle it just fine but the panel is only refreshing at 60Hz. Seiki has promised a firmware update for the 39 inch to fix this as they basically are falsely advertising the 39 inch model as 120Hz right now (as it doesnt upconvert 4k to 120hz at all, and neither does the 50 inch).

I am getting annoyed at how long I have had to wait for this firmware update...

Do you think that they are having issues getting it to work because there's only 2x81 Pin LVDS cables?
 
I would disagree with this on the seiki's... running at 2560x1440 the scaling is horrible it looks like 2560x1440 is getting scaled to 1920x1080 and then back up to 3840x2160 to me.

Hi there,

Do me (and yourself - possibly) a favor, and go to your Nvidia/AMD control panel, and ensure the scaling is done on the GPU.

I'd expect the scaling, and everything in this TV, to be of a lesser quality, where it would shine when you use as little of the "features" of the TV as possible, and let it shine "au natural".

I'm seriously considering this screen, as HDMI 2.0 TVs by budget TV manufacturer's (TCL/SEIKI) don't look to be coming anytime soon. 2560x1440 is probably the only realistic resolution I could use (I want more than 1080p) and 2160p is really too much for today's hardware (for newer AAA games).

Let me know if the 1440p shows a big improvement using the Graphics card scaler, and is the 1440p scaling on the TV really that bad? Where it doesn't actually show a noticeable improvement over 1080p?

Thanks
 
Hi there,

Do me (and yourself - possibly) a favor, and go to your Nvidia/AMD control panel, and ensure the scaling is done on the GPU.

Of course scaling looks 'ok' when its using video card scaling but then its only running at 30Hz and the display will be receiving 4k @ 30Hz so why would you run at 2560x1440 instead of 3840x2160 when its the same refresh rate? I know I would never have a reason to...
 
Of course scaling looks 'ok' when its using video card scaling but then its only running at 30Hz and the display will be receiving 4k @ 30Hz so why would you run at 2560x1440 instead of 3840x2160 when its the same refresh rate? I know I would never have a reason to...

Yeah you're right.. I totally forgot about the fact that the screen would run @ native res @ 30hz...

The only benefit would be Graphics performance, but the 60hz kills it.

So 2560x1440 @ 60hz scaled on the Monitor, looks like sh*t? Would you say it's better than 1080p on the same TV?
 
Yeah you're right.. I totally forgot about the fact that the screen would run @ native res @ 30hz...

The only benefit would be Graphics performance, but the 60hz kills it.

So 2560x1440 @ 60hz scaled on the Monitor, looks like sh*t? Would you say it's better than 1080p on the same TV?


1080p looks significantly better than 1440p. Like I said I am about 90% sure that when you give it 2560x1440 it is getting:

2560 x1440 -> 1920 x1080 -> 3840x2160.

The text is not clear at all the pixels merge together in the text, etc... Here is some pictures (sorry, not the best quality from my iphone:

Just showing the TV is getting 2560x1440 @60Hz:


Close up to terminal and firefox @ 2560x1440. As you can see text is very hard to makeout. It looks better looking at it in person but its definitely the type of image I would expect when something is getting interpolated to lower resolution than the actual image was:



Same thing now at 1920x1080 (1080p scaling, looks way better).



And same thing now at 3840x2160 (looks the best):
 
Even with the phone pics you can see the issue.

Notice how when its 2560x1440 that the text appears to have grey-ish spots (its not as bold/white as the other resolutions) I believe that is because its being aliased down to 1920x1080 so due to it going across sub pixels it makes the text look grey-ish from the aliasing.
 
Yeah, there's a noticeable difference, 1440p is clearly scaled. Thanks for taking these pics btw, helps out with my decision.

Did you ever try 125% scaling in Windows? I know that 1440p @ 125% is the same size as 1080p, which would help determine if text/images are actually sharper/better @ 1440p.

Considering that 1440p on a native 2160p screen is approximate scaling, I wouldn't expect perfection, but I'd still expect better than 1080p images/text.
 
Yeah, there's a noticeable difference, 1440p is clearly scaled. Thanks for taking these pics btw, helps out with my decision.

Did you ever try 125% scaling in Windows? I know that 1440p @ 125% is the same size as 1080p, which would help determine if text/images are actually sharper/better @ 1440p.

Considering that 1440p on a native 2160p screen is approximate scaling, I wouldn't expect perfection, but I'd still expect better than 1080p images/text.

No, I don't have very easy access to windows but honestly I would think if its rendering text at the same physical size that the quality between 1080p and 2560x1440 would be very close because its already being downscaled to 1080p but honestly i don't think this is any better than running 2560x1440 on an actual 1080p TV. I do not recommend this TV at all for 2560x1440 usage. Now the skyworth could be a very different story.

This is of course on the 50 inch version. I can try the 39 inch tomorrow at work.
 
Here is an example of the top part native pixels and the bottom part is being scaled down to 75% of normal for horizontal and vertical resolution and then scaled back up to 100%:




This is a very similar effect to what I am seeing. Thus, the one on the bottom looks like crap.
 
Here is an example of the top part native pixels and the bottom part is being scaled down to 75% of normal for horizontal and vertical resolution and then scaled back up to 100%:




This is a very similar effect to what I am seeing. Thus, the one on the bottom looks like crap.

Wow.. I can't imagine 1440p looks as bad as that. I don't think it's being scaled down then up, it's just that @ 1440p everything is smaller, so things won't be as sharp. I have tried running my 40"/46" 1080p TVs @ 1440p, so I know what that's like, but I'd hope and imagine that 1440p on the Seiki would look at least somewhat better, considering you have extra pixels to work with. Again, if the scaler if pretty much worthless, then, I'd understand that the algorithms for doing 1440p (especially when it isn't even officially supported) would be next to worthless.

Disappointing, I was hoping to use it as a 1440p for desktop/gaming. Once we get a HDMI 2.0 display, we can safely use the Nvidia/AMD scaler, and end up with a better image @ 60hz for 1440p.

I guess I've got to wait it out after all. I hope Seiki or TCL make a 39" with HDMI 2.0 sometime soon. Or perhaps I can learn how to replace the HDMI and TCON and "upgrade" it to HDMI 2.0 if that's not too difficult.
 
Wow.. I can't imagine 1440p looks as bad as that. I don't think it's being scaled down then up, it's just that @ 1440p everything is smaller, so things won't be as sharp.

That would be the reverse. Assuming native pixel sizes than the smaller it gets the *shaper* it gets. Things are sharpest at 4k, then comes every resolution 1920x1080 and below and then finally at the very end 1440p. Every resolution 1920x1080 or under looks better than 1440p.

I used 1440p on monitors for lots of years as well as 4k on a 22 inch monitor. Trust me its downscaling it and upscaling it again incorrectly. It was very obvious to me from the first time I tested the resolution because you can see the downscaled aliasing that is happening in the text as its not pure white but becomes grey-ish in certain areas.

The skyworth panel could be different.

Also another reason I know its downscaling to 1080p is because when you use video card scaling (instead of monitor) it looks about 10x better and like the other resolutions. Of course then its only 30hz so what was the point?
 
Even with the phone pics you can see the issue.

Notice how when its 2560x1440 that the text appears to have grey-ish spots (its not as bold/white as the other resolutions) I believe that is because its being aliased down to 1920x1080 so due to it going across sub pixels it makes the text look grey-ish from the aliasing.
I don't see any point to use 250x1440 as desktop resolution. But for most games, it is OK.
 
I don't see any point to use 250x1440 as desktop resolution. But for most games, it is OK.

In my oppinion even games look better at 1080p than they do at 1440p because of the down scaling its doing and at 1080p you get 120Hz (atleast on the 50 inch one).
 
Here is an example of the top part native pixels and the bottom part is being scaled down to 75% of normal for horizontal and vertical resolution and then scaled back up to 100%:
This is a very similar effect to what I am seeing. Thus, the one on the bottom looks like crap.

If you are trying to simulate the scaling of the monitor, you are producing results that are far inferior to what the monitor is producing.

The way you are doing it, you are throwing away information, but in reality, the scaler doesn't throw away information.

You just need to take your original image and scale it up with a non-linear multiplier.
The scaling method you choose will determine the quality.

In this simulated example from native from 3840x2160 to 2560x1440,
From from top to bottom, sorted by what I think is best to worst:
Original, Bilinear, Nearest Neighbour, Bicubic

image
 
Last edited:
If you are trying to simulate the scaling of the monitor, you are producing results that are far inferior to what the monitor is producing.

The way you are doing it, you are throwing away information, but in reality, the scaler doesn't throw away information.

You just need to take your original image and scale it up with a non-linear multiplier.
The scaling method you choose will determine the quality.

In this simulated example from native from 3840x2160 to 2560x1440,
From from top to bottom, sorted by what I think is best to worst:
Original, Bilinear, Nearest Neighbour, Bicubic

Original%2C%20Bilinear%2C%20Nearest%20Neighbor%2C%20Bicubic.png


[sorry, I know the link is broken, trying to fix it.]


Did you read my posts man? Trust me the monitor *is* throwing away information. It is basically taking 2560x1440 and downscaling it to 1920x1080 and then upscaling it back to 3840x2160. Its obvious because 1440p looks like CRAP on this display. As for 1080p scaling I have definitely seen better but at 1440p it is about 10x worse than video card scaling compared to 1080p which is only what I would call 'slightly' worse than video card scaling.
 
Here. In these pictures its close enough that you can actually see the physical pixels:

Look at how crappy the text is. In the title of this thread the e in 'Seiki' is almost unreadable:




Look on the menu again how horrible the fonts look and stuff is meshed together:



Notice how like all the black text in it has white pixels in it and whatnot? The effect looks exactly how I would expect zooming into something that was downscaled.
 
Did you read my posts man? Trust me the monitor *is* throwing away information. It is basically taking 2560x1440 and downscaling it to 1920x1080 and then upscaling it back to 3840x2160. Its obvious because 1440p looks like CRAP on this display. As for 1080p scaling I have definitely seen better but at 1440p it is about 10x worse than video card scaling compared to 1080p which is only what I would call 'slightly' worse than video card scaling.
It's pretty clear the SEIKI is indeed doing an incorrect 2-pass scaling of 1440p here.

Perhaps it's scaled DOWN to 1080p by one chip (throwing away information from 1440p), then scaled UP by a different chip, e.g. perhaps some kind of an inability to do anything 60Hz at *native* resolution above 1080p.
 
houkouonchi,

Sorry, I didn't read the earlier threads carefully enough.

I understand what you are saying, but it's difficult for me to see what the difference is without a back-to-back. (i.e. I don't know what the original 1:1 looks like, so it's hard for me to tell how far off the "bad" screen shot really is.)

You could create a bitmap of alternating white and black pixels ( a very find checkerboard grid at 1 pixel spacing) and see how it mangles that up. That should very quickly show if it's dropping pixels.

As a side note, my system (7950+39") doesn't even have the option for 1440p.

It has:
3840x2160
2048x1536
1920x1200
. . .
 
I understand what you are saying, but it's difficult for me to see what the difference is without a back-to-back. (i.e. I don't know what the original 1:1 looks like, so it's hard for me to tell how far off the "bad" screen shot really is.)
I have enough familarity with scaling (I developed line doublers for a living more than ten years ago, and I was the inventor of the world's first open-source 3:2 pulldown deinterlacer, found in dScaler in the late 90's). So I can confirm, that the images, are, definitely, indeed, degraded by 2-pass scaling -- just simply by looking at images.
 
I have enough familarity with scaling (I developed line doublers for a living, and I was the inventor of the world's first open-source 3:2 pulldown deinterlacer, found in dScaler in the late 90's). So I can confirm, that the images, are, definitely, indeed, degraded by 2-pass scaling -- just simply by looking at images.

Thanks for the confirmation Mark.
I tried the possible resolutions for my system, but none of them appear to be dropping information like the 1440p examples.
 
You could create a bitmap of alternating white and black pixels ( a very find checkerboard grid at 1 pixel spacing) and see how it mangles that up. That should very quickly show if it's dropping pixels.
Oh -- it could -- but it doesn't necessarily work that way. If it's a bilinear/bicubic/advanced scale algorithm, it simply blurries up the pixels and adds nyquist/moire artifacts, and if you do two passes, the artifacts compound on top of each other, with a doubled-up gleen of softness on top of it all. I recognize the telltale artifacts of a 2-pass scaling on the images. The different passes of scaling may even use different scaling algorithms (one scaling pass point-sampled, second scaling pass bicubic-sampled, as an example), creating different kinds of artifacts.

You can view Lagom LCD test patterns here for the alternating white pixels -- www.lagom.nl/lcd-test
Use the Tracking/Phase pattern as one great example, or the Lagom LCD Inversion pattern.
 
Thanks for the confirmation Mark.
I tried the possible resolutions for my system, but none of them appear to be dropping information like the 1440p examples.
The 39" and 50" may be doing 1440p scaling differently.

As an indication of how the scaling is defective -- either two passes or every-other-dropped -- Look at the text saying "Full screen" in his photograph. The two letter L's in "Full" are blended together. That shouldn't happen like that when upscaling unless it's two-pass or pixel-dropped before scaling. It may even be dropping pixels during the first scaling pass. A resolution test pattern will determine this.
 
The 39" and 50" may be doing 1440p scaling differently.

As an indication of how the scaling is defective -- either two passes or every-other-dropped -- Look at the text saying "Full screen" in his photograph. The two letter L's in "Full" are blended together. That shouldn't happen like that when upscaling unless it's two-pass or pixel-dropped before scaling. It may even be dropping pixels during the first scaling pass. A resolution test pattern will determine this.

For clarity, my system is not doing any 1440p scaling because it's not an available resolution option.

The lagom website is useful. thank-you.
 
I have enough familarity with scaling (I developed line doublers for a living more than ten years ago, and I was the inventor of the world's first open-source 3:2 pulldown deinterlacer, found in dScaler in the late 90's). So I can confirm, that the images, are, definitely, indeed, degraded by 2-pass scaling -- just simply by looking at images.

Same here (as far as being able to tell just by looking at it). Like it was obvious to me within 5 seconds of using that resolution. I have been trying to tell people that its being downscaled to 1080p and then upsaled back to 4k a bunch of times and nobody seems to believe me LOL.

I am at work but busy right now (the 39 inch is sitting in front of me). I will give 2560x1440 a shot later to see if its the same on the 39 inch.
 
stupid question of the day: way would anyone want 1440p on a 4k display?

Scaling is bad, the lesser evil is half resolution scaling, where you simply take for small pixel and use as the building block of your low resolution image in 2x2 or 3x3 blocks.

Examples:
2560x1600>1280x800= clear scaling 2x2 blocks
2560x1600>anything else= blurred scaling

3840x2160>1920x1080= clear scaling 2x2 blocks
3840x2160>1280x720= clear scaling 3x3 blocks
3840x2160> anything else= blurr
 
Back
Top