totally agree, the 8k is a failed product. But my point is only that upscaling is not bad by definition. It works great in the Pimax 4k.
Well as a 4k owner I can say no render multiplying Steam vr SS. The 1440p upscaled does not look as good as render multiplier cranked the image improves overall.
But your right Nvidiaās DSR & Amdās VSR are just gimicks & waste of gpu resources.
As he said you think it does by comparing lcd rgb to pentile.
So both are true. If the 4k can create a decent upscaled image so can the 8k. But if you had a Qhd rgb lcd, this 2.5k headset will look sharper than the 4k model. As shown between 5k+ & 8k.
Native is better overall vs upscaling/resizing frames.
Well obviously not. The pimax 8k fails to deliver a good, sharp, detailed, rich image (like the pimax 4k pics Iāve posted). Youāre one of the very few on the forum who still doesnāt see that. Or at least pretends to not see it.
Not so there are also many whom are sticking with the 8k.
Both @neal_white_iii & @mmorselli have recently posted some interesting info.
And when are ypu going to compare apples?
Rgb vs rgb instead of rgb vs low res pentile.
@Sjef, I think your opinion is related to pixel density. If we look at the information that has been posted on the forums through testers so far, Pimax 8K seems to have a lower pixel density than 4K. Therefore, I think that it is reasonable to doubt the utilization rate of the panel. On the other hand, 5K + seems to have a high pixel density. But if it is due to the size of the 8k panel, weād better expect 8K +.
I would have to agree that a Native resolution signal will ALWAYS look better than an upscaled image, On the same display. Iām not saying an upscaled image will ALWAYS look bad. Sjefdeklerk your a smart guy you must clearly be aware your godly 4k pimax would look even better than it does now with a native signal, right?
Another thing that I just canāt seem to get my head around isā¦ What does the panel utilization have to do with the image clarity when thereās already more pixels in the viewable area on the 8k now, it should already look better than the 5k+ without using more of the panel, if panel utilization was the issue.
From what I understand, the 5k+ has a resolution of 2.5k per panel right? And the 8k has 4k per panel and less visible screen door, right? They both use the same FOV, and lenses set at the same fl, right? if all those statements are trueā¦ it must be a degradation in the signal . Is this assumption incorrect? please enlighten me.
2 Likes
Thanks, thatās very useful! Thank you again @mixedrealityTV!
1 Like
We already know 5k+ looks better than 8k, and thanks to @mixedrealityTV we also know itās the case with max SS.
The problem is that we still donāt know why and if itās something that can be fixed in the future by tweaking 8k scaler. @deletedpimaxrep1 can you try to find out? Cam this be changed by the driver or a firmware update?
@Sjef why do you think 8k canāt be fixed? We donāt know whatās inside the 8k and how configurable it is. Those chromatic aberrations are clearly due to scaling algorithm, itās doing some kind of averages that ruins the final image. If think that it should just duplicate pixels instead of changing/averaging their colours.
For me the scaler is working clearly wrong, is not filling the pixels with the indicated color and it is possible that an update is needed, that instead of interpolating colors filling without more until the algorithm is improved.
When passed the signal 1440p. The following should be programmed:
[original red pixel] [creator pixel by the sccaler] [red pixel original]
The pixel created is red, instead of white, or black as it seems in some photos.
[original red pixel] [pixel creator by the scaler] [blue pixel original]
I would always program the same in case of doubt on the right and ready
Pixel blue and ready.
I would do it in a straight horizontal line, not in oblique lines.
You make it appear as if the improvement of the 8K is not going to happen because Pimax will shift their efforts to the 8K(x).
Actually I believe exactly because the 8K does not seem to fully unleash its potential today, Pimax will not go on a separate development path of the 8K(x) and leave the 8K for what it is, but will simply work on the 8K to improve it and see if it can include the native resolution option in a next iteration of the 8K. That would then marry up the two separate product lines of 8K and 8K(x) into one, possibly dubbed the 8K+. For those who receive the 8K in the coming few months, it of course will not make their units much better, at least unless Pimax find solutions which solely rely on modifications to the firmware & software. For this 8K+ as descirbed above you would definitely need to modify the hardware too, because we need the bandwidth for the native 2x4K signals and the option to bypass the upscaler (and possibly a different upscaler and/or bridge chip(s)).
So I hope that Pimax is just going to improve the 8K and the 8K(x) backers will receive such improved 8K+ next year. That would focus their development capacities on their (next) flaggship product rather than a small quantity side orderā¦
However, it would be pretty crucial to be able to have the option to feed the native resolution, because that was the whole idea behind the 8K(x), and you would see many backers complain if that werenāt in it. Even if the same backers may not be able to put that native resolution to any greater practical use in 2019/2020.
1 Like
Thank to @SweViver we now know that PieTool updates can improve 8k image:
http://community.openmr.ai/t/mrtv-pimax-8k-vs-pimax-5k-plus-sde-quality-comparison-at-maximum-quality-settings-supersampling-on/8394/165
And from below screen we can see what the scaler is doing wrong:
It seems to be using checkerboard scaling. Iām sure that if it was just duplicating pixels the result would be better.
2 Likes
Thank you very much, I knew something was wrong but I could not prove it or get proof.
Thatās not the scaling, but the pixel arrangement of the LED displays.
3 Likes
Youāre probably right, so it means zoom levels are different or pixel density is much lower than Iād have expected on 8k. In any case scaler is doing a terrible job, there should be no blue/red pixels and middle of the line should be bright white.
I believe that the point here is that there are clearly black pixels (or holes) surrounded by white pixels in 8K image.
If the original image contained one solid line, even jagged one, any upsampling should again create a solid object. In 5K+ picture it is solid, in 8K picture it has holes in it.
I could accept holes at the line edge, but not in the interior of the line. At least not with the right scaler.
The worry I think most have is:
Pimax will fix the 8K somehow after people have switched to the 5K+
But with no info from Pimax @deletedpimaxrep1! all people have to go on is the Tester videos that show the problem but not why it exists!
It is very important for Pimax to let people know if this is a software bug (possible fix) or a hardware fault (no fix)
18 Likes
yes, i dont want to change to 5k+ and one month later they fix it the 8k. backers will be very upset. @deletedpimaxrep1 we need some official answers
6 Likes
Take a good magnifying glass and look at your monitor. Youāll be surprised.
3 Likes
When I look at swevivers zoomed in image of the white lines in the steam vr enviroment it looks to me like the problem is two things.
-
the pixel arrangement is different, diagonal pattern on 8k, vertical/horizontal on the 5k+ also the 5k+ pixels seem to mostly touch on the horizontal plane so you only really see black gaps between each horizontal lines. (on the digonal lines I see 5 white pixels per line separated by black space on 8k and 3 white pixels per line on the 5k+ that touch each other horizontally). This leads to any horizontal line looking perfectly sharp on the 5k+ while broken up on the 8k (Rephrased the 5k+ pixel pattern gives near perfect pixel fill ratio on horizontal lines while 8k canāt). Nothing can ever be done about this. I donāt think there is any actual pixels showing black that shouldnāt be, I think there is simply more inter pixel black space showing/lower horizontal pixel fill ratio on 8k. The only way to make this match on 8k would be to fit new panels that use the same pixel arrangement as 5k+.
-
The brightness of what should be a solid white line drops towards the edges. it doesnāt seem to actually blur the detail over more pixels than it should be but the edge contrast is not maintained which makes it very hard for human eyes to see the edges. Itās possible that an edge sharpening filter somewhere in the render pipeline could largely mitigate this for a much smaller overhead than extra supersampling. Iām not sure however if the filter would need to be post upscale in which case it would need to be implemented on the headset which is probably not possible at this stage. If however doing this sharpen pre lens warp or pre transmit over hdmi before the upscale works then this might well end up being a solution pimax can and do implement. Itās unlikely to ever fully match the 5k+ but I would not be surprised if it gets pretty close even at pitool render setting of 1.0. Might not ever happen though.
2 Likes