Orlando Pimax Roadshow Impressions (8k X, 8k+, 5k Super)

Awesome impressions and pics! Thanks added to spotlight.

3 Likes

Did anybody ask about backporting of the new distortion compensation to the older models?

( I heard the “we had to redo from scratch because of smaller display panels” -line repeated, but that does not really wash – going to need a more specific technical description. )

Apparently the poor Artisan on hand lay forlorn on its side table, but did anybody give it enough attention to get an impression of how it deals with IPD? -Are the 9:8 screens wide enough to accommodate shifting the image for IPD, whilst retaining full vertical resolution? …or does the FOV you get hinge on your IPD? …or do we have such a major change, that the screens now move with the lenses, with this model?

1 Like

As you rightly pointed out distortion profiles appear to be custom made for the panel size. It’s a bit like asking if you could backport my eye glasses to you?

Maybe they could redo from scratch the older model distortion profiles using the knowledge they have learned from the Vision series? Whether they would want to do that for older models is a different question :blush:

2 Likes

Did you tried 8K+ …are there difference between them in game not SIM?

1 Like

This I think could be confusing statement. So will give my take on it.

Distortion profile itself may still be at the core the same.

Now if we take a look at say the 5k+ vs the unseen 5k. The 5k presumably originally had the same physical size Panel as the 8k. If mem serves from fellow whom changed 1 of his og 8k 4k panels after breaking one dropping the headset? It was a 5.5" panel? Where as the 5k+ if mem serves was a 5" panel from Sf’s teardown.

So now 5" panel has greater ppi vs a 5.5" panel.

So for simplicity lets use:

Panel (A) has a ppi of 100 and Panel (B) has a ppi of 120 (Panel A & B have same resolution)

This changes non only X and Y coordinates of lense distortion but Panel (B) also has more pixels to refine the distortion.

So 8k+ & X being presumably smaller panels than the og 8k and better quality panel would need a reworking of the distortion algorithm. Plus the Vision 8ks having rgb on top of this.

2 Likes

I am not quite sure what you are saying here - I could read the above in a couple different, contradicting, ways.

The distortion nature of a profile has zero to do with the panel size, which plays in only as a matter of scale, and one that will already have had to be dealt with before, given the size difference between the 8k and 5k+. -Granted: It could have been implemented exceedingly badly, and hardcoded, but profile data we have previously seen inside earlier versions of piserver suggests sensible scalability.

All we have heard so far, suggests that the new devices have the exact same lenses, mounted the exact same way, as the 8k and 5k+, and that is the determining factor; The physical optical properties, that the software predistortion compensates for; I do not need different strength glasses to read a small book with 8pt print at armlength, to a large one with 8pt print at armlength.

Now: We could very well be dealing with an entirely new algorithm here, instead of “just” a nicer profile for the old one, but that still doesn’t change the optical hardware, which is, as far as we know, common between the old and the new devices.

Not really, unless the original algoritm was very, very badly designed and coded, and if that were the case, I wouldn’t rule out that if would have forced the game to render at a certain resolution, too, without allowing for supersampling.

The distortion would by any measure of reason be applied over a surface at a size and resolution that is given in the profile, and scaled appropriately, using floating point math, making it fit any bitmap -size, -resolution, and -coverage pattern.

Does the old algo, or any hypothetical new one, do subpixel-layout-aware rendering, for better spatial accuracy? I seriously doubt it - especially given we have the 8k scaler butting in and doing its thing, before imagery goes up on the screens. :7

1 Like

Sorry- was offline- sure might make sense to pull it together!

I actually didn’t even try the 8K+, I had preordered the 8KX and only focused on that.

4 Likes

That’s because it is. Angular resolution of the 8kX is theoretically 30 pixels/deg, theoretically about half that of 20/20 human vision, equal to that of 20/20 human vision with artifacts, and equal to that of typical desktop monitor setups.

2 Likes

Most definitely. It’s really obvious. It’s like a 6 or 7. The bump up in sharpness is super noticeable. In Beat Saber, the 8k X felt like a monitor in terms of sharpness.

4 Likes

Oh yea, here’s something interesting. IIRC I heard that a company came to Pimax and showed them real time distortion correction. It only affected fps by one. Hopefully Pimax will do all they can to get this implemented into Pitool. It’d be a pretty great feature for the eye tracking.

3 Likes

Assuming that 8k-X has the same panel size as P5k+ and the same optics (lenses) and considering that P5k+ has ~ 14 PPD (https://community.openmr.ai/t/pixel-per-degree-ppd-of-pimax-5k/19389), it would give ~ 21 PPD. How did you get 30?

1 Like

Pretty sure it’s smaller. Not sure where it was mentioned but I believe it’s 5" instead of 5.5" on the 5K+ and that’s why they had to redo the distortion profiles (which turned out better than on the 5K+ apparently).

6 Likes

I thought 5.5" was the original 8K and 5" was 5K+.

1 Like

It’s a rough calculation intended mostly for rough comparison purposes with an estimated +/-40% error. I don’t remember the exact formula at the moment, but it is included on my param spreadsheet.

When you wrote theoretically I assumed you have some theory behind to support the claimed value. “Rough comparison with +/- 40% error” could be interesting for you, but I would not post it on the forum as a fact without the big disclaimer.

3 Likes

That’s how I recall it.

If the 5K Super has reduced distortion, it’s pretty safe to assume it could be applied to the 5K+. Have we had any comments about the 5KS distortion?

1 Like

I felt there was much more reason to be confident in the 30pixel/deg number for the Pimax 8kX than various numbers for other headsets, actually. IIRC, the calculation was basically vertical FOV divided by vertical pixels.

What fov is used, normal or large?