Comparison of VR headsets' displays

[apollon01]

This post got significantly edited on 4 August 2019. The original version is kept below for reference only.

 

Well, after reading through a number of discussions related to FOVs, PPDs, etc. I have to admit I have underestimated the complexity of this issue. Therefore this post is completely re-edited to reflect the most accurate data and methodology "out there".

 

Notes to the methodology and sources:

- All values and calculations are per one eye only (therefore the FOVs are not for the whole HMD but for one eye only).

- The calculations are done for horizontal as well as vertical FOVs.

- FOVs are rendered FOVs, so it is an upper limit of what can be seen (i.e. Valve Index's value is similar to Vive Pro's but the Index allows to see more of the rendered FOV than the Vive).

- Rendered FOVs come from a function in OpenVR IVRSystem::GetProjectionRaw which returns the tangents of the angles of the left, right, bottom and top clipping planes for each eye projection. Refer to risa2000 post on pimax forum for further details.

- Horizontal and vertical rendered FOVs values are sources from a table created by Durante using the OpenVR function above (reddit thread also here). HP Reverb values were sourced from one user on Mudspike forum.

- I am aware of the ROV room in SteamVR that allows users to measure various qualities of their HMDs. After having read through many discussions and having tried it myself, I decided to disregard this method of assessing the visible FOV in favor of the rendered FOV methodology instead.

- For Pimax, the vertical rendered FOVs is with parallel projection off.

 

Known limitations:

- Panel utilization factor not accounted for.

- Image warping and optics effects not accounted for (check this thread for more details; nevertheless, the vertical PPD of the Pimax 5k+ in the thread is very close to the one in the table below; not so for the horizontal one though).

 

And here is the resulting table:

 

Please do read the discussion below for potential further limiting elements and full understanding of the changes made to the original post. I also recommend to read the discussions linked in this post for better understanding of the topic.

 

Thanks all for contribution. I believe it led to a better result (=table).

 

Milan

 

 

---------- Original post ----------

 

Since time to time there are topics about comparison of this HMD to that HMD etc, I made a little table comparing some of the attributes of the displays of the most used VR headsets (my subjective selection). The main goal was to compare PPD - pixels-per-degree of the displays.

 

The table together with the methodology is published on Mudspike forum here:

https://forums.mudspike.com/t/vr-headsets-display-comparison/8756

(There is a number of VR savvy guys there so it is well worth reading and checking once in a while)

 

For your convenience and ease of reading, I also attach the table:

 

For a short explanation of Pentile vs RGB arrangements, check this post here on Il2 forum:

 

My intention is to keep the table up-to-date once new headset are released.

 

I hope you find it useful

 

Milan

Edited August 4, 2019 by apollon01

[Alonzo]

I think you should be careful with the FOV column. Rift S is arguably slightly narrower FOV than CV1, and Chili has measured the Index horizontal FOV at 108 degrees not 130. Similarly, people argue about the Reverb FOV. Other than that I like the table, especially the critical "subpixels per degree" measurement.

[chiliwili69]

I think you should include both horizontal and vertical FOVs in your table from people who actually has measured that, not with the numbers provided by the manufacturers (they intentionally merges terms and you never know if they refer to diagonal or horizontal, etc).

 

There is an application to measure both vertical and horizontal FOVs, among many other things: TestHMD

One important aspect here is that he visualized FOV in any headset can not be larger than the rendered FOV which is reported by the headset driver and calculated in this table.

So Rift-S never can have an horizontal FOV per eye larger than 88º. 

So the rendered FOV would be the maximum achievable FOV of every headset. So you must correct your table a bit.

 

Once the FOV is know, the PPD is not as simple as you may think to calculate it.

Firstly, you need to realize that there is less pixels per degree in the center than in the sides of the panel:

 

Secondly, you need to take into account the panel utilization factor which varies greatly from headset to headset. (for example 70% in Pimax5K+).

 

But despite of this, the real PPD calculation is a bit more complex:

https://forum.pimaxvr.com/t/ppd-discussions/19762

https://forum.pimaxvr.com/t/pixel-per-degree-ppd-of-pimax-5k/19389

 

So, the info of your tables are a bit misleading for people looking for a new headset.

Edited July 25, 2019 by chiliwili69

[apollon01]

Hi @chiliwili69 and thanks for the feedback!

 

There are many valid points in your post. I am aware it is not perfect and as I mentioned on the mudspike forum, it is not meant to be rocket science  but a quick comparison.

 

On 7/22/2019 at 9:10 PM, chiliwili69 said:

I think you should include both horizontal FOV in your table from people who actually has measured that, not with the numbers provided by the manufacturers (they intentionally merges terms and you never know if they refer to diagonal or horizontal, etc).

Sure, however the percieved FOV changes with variables like IPD or eye distance to the display. So even user measurements would have to be taken with a pinch of salt.

 

On 7/22/2019 at 9:10 PM, chiliwili69 said:

calculated in this table.

Valuable source. Thanks for it.

 

On 7/22/2019 at 9:10 PM, chiliwili69 said:

you need to realize that there is less pixels per degree in the center than in the sides of the panel

True for the whole display. However in my opinion the numbers are still comparable because this phenomenon is applicable to all of them (maybe in a different manner to Pimax HMDs…).

 

On 7/22/2019 at 9:10 PM, chiliwili69 said:

panel utilization factor

Thanks for the comment. This is however Chinese for me

 

On 7/22/2019 at 9:10 PM, chiliwili69 said:

So, the info of your tables are a bit misleading for people looking for a new headset

With all my respect, I do not think so. The table is ment to be used as a relative comparison of the HMDs. So I suppose that even if all the above was actually taken into account, the relative ranking would not change too much.

 

Again, it is not science. Just a quick comparison using rather a simple and understandable methodology.

 

Thanks for your contribution in any case. Well appreciated. I will correct my table once I find a minute.

 

Milan

Edited July 24, 2019 by apollon01

[TCW_Brzi_Joe]

14 hours ago, apollon01 said:

Panel utilisation...

You probbably found it by now, but it is good to explain for others; as I understand, the panel utilisation is percentage of HMD screen visible to eye. All HMD panels are rectangular, but we see more or less oval picture. That means that some border pixels are useless. They may or may not be rendered, but either way we payed something that we do not have.
 

SweViver made one good explanation right on this link.

[SCG_Fenris_Wolf]

On 7/15/2019 at 9:38 PM, apollon01 said:

Since time to time there are topics about comparison of this HMD to that HMD etc, I made a little table comparing some of the attributes of the displays of the most used VR headsets (my subjective selection). The main goal was to compare PPD - pixels-per-degree of the displays.

 

The table together with the methodology is published on Mudspike forum here:

https://forums.mudspike.com/t/vr-headsets-display-comparison/8756

(There is a number of VR savvy guys there so it is well worth reading and checking once in a while)

 

For your convenience and ease of reading, I also attach the table:

 

For a short explanation of Pentile vs RGB arrangements, check this post here on Il2 forum:

 

My intention is to keep the table up-to-date once new headset are released.

 

I hope you find it useful

 

Milan

Mate, the FOV (horizontal) types are false, and also ignore the vertical FOV.

 

Index has a diagonal FOV of 130°, not horizontal.

The Reverb's diagonal FOV is 114°, not horizontal.

The Odyssey reached 105° horizontal FOV.

Both Pimax (all versions) and Index have 105° vertical, larger than all the other headsets.

..etc.

I suggest you only stick to diagonal values to incorporate the higher vertical FOV of Index and Pimax, or separate horizontal/vertical and fix the FOV values in the table.

 

The only horizontal values that are correct in this table are Rift CV1's and the Pimax'. The rest are all diagonal values.

 

Also, the PPD of the Pimaxes are similar to the Index, they aren't that much better, because of the panel utilization. The PPD Subpixels of the Index are also too low in the table.

 

P.S. in SteamVR there is a Home Room to visit called ROV. It lets you enter a test environment where you can check all the values yourself. It's free.

P.P.S. Will change my reaction if the tables show correct data ?

Edited July 26, 2019 by SCG_Fenris_Wolf

[apollon01]

Hi @SCG_Fenris_Wolf and thanks for your contribution.

 

I am already working on the new version taking into account your and @chiliwili69 comments. As a new source, I will use the table chiliwili69 quoted (and will try to work on diagonal dimension rather than on horizontal only - in order to capture "square degree" if you see what I mean).

 

However…

...using that table, I am not able to reconcile your claims:

23 hours ago, SCG_Fenris_Wolf said:

Index has a diagonal FOV of 130°, not horizontal.

The Reverb's diagonal FOV is 114°, not horizontal.

The Odyssey reached 105° horizontal FOV.

Both Pimax (all versions) and Index have 105° vertical, larger than all the other headsets

 

Index seems to have diagonal FOV of 114.1°.

Reverb is not included

Odyssey has 97.4° horizontal FOV.

There are many other HMDs with higher vertical FOV than Pimax and Index.

 

Any insights why the table differs from your numbers?

(I suppose - reading this post - that horizontal FOV is… well… horizontal FOV, whereas total hor. FOV is diagonal FOV)

 

Thanks for comments.

 

Milan

 

PS: Keep the reaction the way you see fit. The aim of the post is not to collect reactions

Edited July 27, 2019 by apollon01

[SCG_Fenris_Wolf]

The numbers you just posted are wrong too. Index does not have a diagonal FOV of 114.1°, while having a horizontal FOV of 130°. That's smaller - that's simply impossible. Basic geometry. Whether you want to continue this, without actually having tested the headsets, and understanding what you're posting, you may want to stop trying to build a table. The reason is that it misdirects people. I see no way how I can be more polite regarding this, I don't mean no harm when I say this, honestly. Good luck to you anyway.

 

Best regards

Edited July 28, 2019 by SCG_Fenris_Wolf

[TCW_Brzi_Joe]

8 hours ago, SCG_Fenris_Wolf said:

...you may want to stop trying to build a table. The reason is that it misdirects people....

I totaly disagree. Even if some parts of table may be wrong, or something else important is not included (i.e. that panel utlisation, or unmesurable lences quality) the table gives a lot of data on one place. Take this sheet as a fast comparable chart of different panels, and rough fov, and also how many subpixels are actually there. 

 We all are different, and have different wishes in vr. Somebody likes biggest possible fov, and I like rather bigger density of pixels in lower fov. Somebody else may use any HMD, but for me my ipd 72-74mm reduce a lot what can I use.

@apollon01 please carry on with your work, even if something is unclear or even false. VR production is not a one time thing, it is developing, and changing. I also change my mind about what would be ideally for me, and I change it every few months I was dieing about pimax 5k+, and now I dream about XTAL. Who knows what is next

[SCG_Fenris_Wolf]

You may disagree of course, but among your reasons you didn't pick up the essential problem. False data. It's not just some parts.

 

And as you said, "take this sheet as a fast comparable chart of different panels and rough fov, and also how many subpixels are actually there" - these are actually the false rows. So you make a good case example of being (unintentionally, without doubt) misled. So people sponge false information up easily without actually checking.

 

 

I don't doubt his good intentions, but then he must make it right or not do it. It's his responsibility to not post false data.

Edited July 28, 2019 by SCG_Fenris_Wolf

[apollon01]

Guys, let's keep the discussion civil and emotion-free. After all no one here has bad intentions.

 

Above, I acknowledged that my table was not perfect and commited to correct it time and good data permitting.

 

Therefore I appreciated chiliwili69's reply since it brought new data and evidence on the table. I will work with these (as I mentioned above).

 

Before I make a new version, I just want to cross check the data. That's why I kindly asked @SCG_Fenris_Wolf about his opinion regarding the difference between the table referred to by chiliwili69 (that shows rendered FOV using coherent methodology across a number of HMDs).

 

@SCG_Fenris_Wolf, would you mind to disclose your source / evidence that the Index has 130° diagonal FOV? The product page does not indicate any precise number and the rendered FOV table (which looks credible to mee) is not confirming your claims. On different fora I saw different (of course conflicting) claims including one saying that Index actually renders similar FOV to the Vive. The rest is done by the lenses.

 

Thank you.

 

Milan

 

 

[apollon01]

Well, in the end the difference might be explained by the rendered FOV vs visible FOV (hence the 114.1 vs 130 for the Index).

 

So I can imagine updating the overview with correct rendered vertical and rendered horizontal FOVs however with the limitations that the comparison would not take into account:

- panel utilization

- different stereoscopic overlaps among HMDs

- difference between rendered and visible FOVs (due to lenses "magic" and the optics staff in general).

 

It is unlikely I would take it farther than that with the data, knowledge and time I have.

 

I am grateful for your opinions.

 

Milan

 

 

Edited July 28, 2019 by apollon01

[skline00]

appollon1, thank you for all the work you did to compile this table.

 

I own the Oculus CV1 and just bought the Oculus Rift S as a birthday gift for my grandson.

 

The clarity of the Rift S is amazing.

 

He plays a ton of VR so the Rift S will be put to good use.

 

Thanks again.

[SCG_Fenris_Wolf]

On 7/28/2019 at 2:21 PM, apollon01 said:

Guys, let's keep the discussion civil and emotion-free. After all no one here has bad intentions.

 

Above, I acknowledged that my table was not perfect and commited to correct it time and good data permitting.

 

Therefore I appreciated chiliwili69's reply since it brought new data and evidence on the table. I will work with these (as I mentioned above).

 

Before I make a new version, I just want to cross check the data. That's why I kindly asked @SCG_Fenris_Wolf about his opinion regarding the difference between the table referred to by chiliwili69 (that shows rendered FOV using coherent methodology across a number of HMDs).

 

@SCG_Fenris_Wolf, would you mind to disclose your source / evidence that the Index has 130° diagonal FOV? The product page does not indicate any precise number and the rendered FOV table (which looks credible to mee) is not confirming your claims. On different fora I saw different (of course conflicting) claims including one saying that Index actually renders similar FOV to the Vive. The rest is done by the lenses.

 

Thank you.

 

Milan

 

 

I tested it myself.

 

How to:

Put on Index headset. Start steamvr room ROV. Same with Pimax and Samsung Odyssey. Repeat with additional headsets if you will. 

It allows you to test vertical and horizontal FOV. You can either then calculate diagonal, or have it easy and tilt your head to check diagonal as well. 

 

Also, the PPI of the Index in your table is too low. This has been calculated a few times. It's higher than in your table. It's almost at Pimax5K+ level horizontally, and higher in vertically.

 

To correct the diagonals; Index 130, Pimax 200, Reverb 114, Rift CV1 110. Horizontal: Index 115, Pimax 170, Reverb 100, Rift CV1 95 iirc. 

Vertical: Index 105, Pumas 105, Reverb& RiftCV1 don't remember from top of my head. Will update post in the evening when home. 

 

You can make a table and help people, but as you said yourself,  correct data would be important. 

 

 

P.S. Of the quoted table,  check what line 30 reads please. 

 

It says "this refers to what is being rendered,  not what is visible ".

 

I am referring to what the user gets, the actual output.

Edited July 30, 2019 by SCG_Fenris_Wolf

[apollon01]

Original post completely re-edited (as of today 4 August 2019).

 

Thank you all for contribution.

 

Milan