Jump to content

Resolution tests: Rift vs. VivePro (updated with camera through lenses pics)


Recommended Posts

chiliwili69

Being a Rift user, the main reason to buy the VivePro was to have a bump in resolution.

As explained in another post (https://forum.il2sturmovik.com/topic/36162-vivepro-vs-rift-for-il2/) it was not the main but the only reason. So, it HAS TO BE a good reason to spend that amount of money.

Since our brains always try to subconsciously confirm our expected thinkning, I have conducted a series of tests for ID, Spot and Gauge detail in the Rift and VivePro.

 

The procedure for the tests has been (anyone can repeat the experiment with two headsets):

 

1. I have created two test tracks (ID and Spot+Gauge) that can be downloaded from here ( https://drive.google.com/open?id=1TwlweEr6BDtSUwLjxUg_EhPTlntMoy3j)

2. Using Oculus Mirror for Rift or SteamVR Display Mirror for the VivePro in a 4K monitor, I have just launch the test track and make a screenshot before unpause (so it is the same image for everyone)

3. Then I saved than image has a .png file to not loose any resolution

4. Then I open those image, magnify for 300% or 500%, and then make the composition pictures shown below and save it in .png file.

 

I have also made the test with different SS values in SteamVR in a way that they render the same amount of pixels:

Using SS=100% in Rift is equivalent to use SS=48% in VivePro

Using SS=210% in Rift is equivalent to use SS=100% in VivePro

Using SS=300% in Rift is equivalent to use SS=140% in VivePro

The table is this one:

5af86f5b45153_Testsresolution.jpg.b734a6ca99a2cf78db2b5887b57125d4.jpg

 

1. IDENTIFICATION TEST:

5af8703852e70_IDRift_vs.VivePro.png.08caa5d25218f7300b47377d027283da.png

 

2. SPOTTING TEST:

5af8704f21973_SpotRift_vs.VivePro.png.c63c14cc06f4590bf75e40074fce04c0.png

 

3. GAUGE TEST:

5af870674e750_GaugeRift_vs.VivePro.thumb.png.97983b91f9e41fe8b7b7a276240531d4.png

 

Here I would like to make some comments about:

 

1. This images are taking from Oculus Mirror and from Display Mirror. They are the images after the render and before applying the lens distortion correction (it is the inverse of what the lens will do to the image on the display). This correction stretchs (or compress in the center) the image differently in Rift and VivePro. In this article you can look how the image is transformed in Rift and original Vive: http://doc-ok.org/?p=1694

 (Fig 5 and Fig 10):

For Rift:

DistortionMeshRift.thumb.png.0f89125cd56e41be05222c00aca4fe7d.png

 

For Vive (I don´t know if this is the same used in VivePro):

DistortionMeshVive.thumb.png.70c3b0fcb314df52c76d96946b855c74.png

 

2. As commented in the another post (https://forum.il2sturmovik.com/topic/36162-vivepro-vs-rift-for-il2/?do=findComment&comment=615505) the image rendered by the VivePro is much larger (bigger FOV) than the one used in the Rift. But afterwards, you see the same view in the VivePro than in the Rift. So many pixels in the VivePro are not as effectively used as in the Rift.

 

 

 

3. I also obtained some images through the VivePro and Rift with a mobile phone camera. I picked the area of the He-111 and compare Rift and VivePro. What it surprises me is that the image of the VivePro seems to be much better in the camera picture. So the image displayed by SteamVR Display Mirror might not be the real image used by the device. If this is true (which is what it seems) all these comparisons has no sense (:huh:). The only way to effectively compare the devices would be using a proper camera situated at the right point in the lens of the device:

 

5af87dc3a6599_Cameravsmirror.png.d3a025ada81d70db639aec74473636e1.png

 

Updated 17-May-2018: See more picture with camera below

Edited by chiliwili69
  • Thanks 4
Link to post
Share on other sites

I appreciate the time and effort put into these tests but as you mention yourself in point 3, the fact you are relying on what is rendered to the Steam Mirror makes it mostly moot IMHO.

 

As you correctly say, when playing in VR we don't use the Steam Mirror app , we use our HMDs.  Only your tiny through the lens images convey this in any meaningful way.

 

This guy did a really good video through the lens of the Vive, CV1, Lenovo Explorer and Vive Pro.  I have linked from near the end of the video to the Elite Dangerous section where we can compare the full screen images with HUD.  We are not talking massive improvement but a lot of the gauges and HUD text becomes clearer on the Lenovo and Vive Pro.

 

https://youtu.be/6Co-SFIWwew?t=5m46s

 

Of course looking at a 3D image on a 2D monitor also brings its own problems.

 

Edited by ICDP
Link to post
Share on other sites
SCG_motoadve

I have both and the resolution of the Vive Pro is much much better, but the sweet spot is so narrow you end up seeing lots of blurries when playing, it is really bad.

 

So at the end you gain resolution , but the experience is much worse because goes from sharp to blurry all the time during game, and when chasing an enemy and shooting, its very annoying to have those changes in resolution.

Link to post
Share on other sites
chiliwili69
12 hours ago, ICDP said:

We are not talking massive improvement but a lot of the gauges and HUD text becomes clearer on the Lenovo and Vive Pro.

 

https://youtu.be/6Co-SFIWwew?t=5m46s

 

 

That´s a nice video. But I don´t know what SS values he used for that comparison. If he used 100% for all then it is not fair. He should use the same number of rendered pixel (so same demand from GPU) in order to be fair.

 

12 hours ago, ICDP said:

As you correctly say, when playing in VR we don't use the Steam Mirror app , we use our HMDs.  Only your tiny through the lens images convey this in any meaningful way.

 

Yes, something should be wrong with SteamVR Display Mirror tool. Maybe it is time to use a GoPro or eye-fish type camera through the lense. Any idea about what it could be a good camera to capture good images through VR devices?

Link to post
Share on other sites
8 hours ago, chiliwili69 said:

 

That´s a nice video. But I don´t know what SS values he used for that comparison. If he used 100% for all then it is not fair. He should use the same number of rendered pixel (so same demand from GPU) in order to be fair.

 

The question is which HMD has the best resolution in OOB configuration, not on relative FPS performance.  If you increase SS on the CV1 to match pixel count on the other HMDs then you are giving the CV1 an artificial IQ advantage.  The only true way to test is from a common baseline which is OOB (out of box).  If we start increasing or decreasing settings from default to meet some illogical and arbitrarily chosen "fair baseline" number then we are introducing unintentional bias.

 

Think of it this way, if we were to reduce the SS on the Odyssey to match the CV1 render target, it would require supersampling to be set at 0.8%.  Obviously this reduces default resolution by 20%, thus rendering (pun intended) any comparison invalid.

 

If we are comparing a 1080p monitor to a 1440p monitor for sharpness and clarity of display, would you turn the 1440p monitor down to 1080p to make the test "fair"?  Of course not, because that would make the test totally invalid.

Link to post
Share on other sites
WIS-Redcoat

I am not surprised by his findings.  As a VivePro user, the only big difference I see is MUCH clearer menu text but alas, a narrow "sweet spot".  I also wonder about this notion that the VivePro IPD is not actually 100% but renders things at 80-90%.

 

 

Link to post
Share on other sites
chiliwili69
12 hours ago, ICDP said:

The only true way to test is from a common baseline which is OOB (out of box).  If we start increasing or decreasing settings from default to meet some illogical and arbitrarily chosen "fair baseline" number then we are introducing unintentional bias.

 

Yes and No.

If the comparison is for general purpose use or watching movies, etc the out of the box settings would be the right to use.

But my comparison was not for general purpose. I made it specifically for IL-2 and for the settings that I would use with Rift and VivePro. Which are the setting that really matter at the end.

When I use the Rift, I can set the SS in the range 170% to 250% with a 1080Ti, (lately I used 200%).

With the VivePro, I set the SS in the 100% to 130%, normally at 100% since not noticeable differences are found.

 

Typically most of us set the SS value to load our GPU at 70-80% at normal play game, so if I make a comparison using SS=100% in both Rift and VivePro, I will not be comparing apples with apples, but unrealistic settings.

 

I was trying (unsuccesfully) to make that comparison to objectively decide If I keep the Rift or I keep the VivePro. So I used the seetings which load the GPU equally (not arbitrarily), which is what I consider IMHO "fair".

 

 

Link to post
Share on other sites
chiliwili69

Well, since SteamVR Display Mirror showed to be not reliable, I have taken some pictures with a compact camera (Panasonic DMC-ZX3 Lumix) with 14 Megapixel images.

The camera autofocus works OK for most of the pictures but in the case of the Rift appear a chromatic aberration which doesn´t exist when looking with a human eye. So when looking the pictures, please, don´t take that into account. Look only for details of the pics.

 

I have also uploaded all the original pictures in this link and you can download from below link, so it is easier to compare them with an image viewer and pressing the arrow key:

 

https://drive.google.com/open?id=1nF8-b9J-R8_hC0vnZMPvgu9WFTfHt93B

 

I have zoomed some areas of the pictures so you can have an idea, but the best is to donwload the collected pictures of above link:

Hangar.png.50a72f0df953206205db3af8a50f1e93.png

 

5afcb017c66c2_IDSpot.thumb.png.a43fea0a4c9d6c1a279bc7a2166152d8.png

 

Gauge.thumb.png.30c06ba75e7719d3f8562e8147ebdf21.png

 

So, after looking the pictures you can see that the pictures are a better in the VivePro (we are rendering same amount of pixels using the settings I would use with the headsets).

But they are not 77% better (Vive display has 77% more pixels than Rift in the display), which is what I was somehow expecting.

The reason for that is that the VivePro waste many pixels rendering an scene that later is not displayed totally in the physical device, so the effective used pixels are only a bit better than in the Rift.

 

On top of that, in the VivePro you have  worse ergonomics, worse headphones, much more smaller sweet spot (the sweet spot of the Rift is huge comparing with VivePro) and a much higher price.

 

With all this I had enough and I am going to sell my VivePro. :huh:

The lens mod (GearVR mod) will only correct the sweet spot, but will not give me the hidden pixels which are not shown.

Anyhow, in few months (or maybe a bit more) should arrive my Pimax8K, so meanwhile I can enjoy the awesome Rift!

 

Link to post
Share on other sites

Excellent images Chiliwili, thanks for posting  Unfortunately the 77% extra pixels thing does not = 77% extra image quality.  The PPD on CV1 vs Vive Pro works out about 35% extra.

 

CV1: 456

Vive Pro: 615 (ish)

 

So it is a noticeable improvement in clarity, as your images show but by around 30%(ish).  One thing that is worth mentioning is the relative performance, on CV1 you need to use considerably more SuperSampling than on a Vive Pro.  Yet it will always have less clarity because of the increased SDE.

 

 

 

Edited by ICDP
Link to post
Share on other sites
chiliwili69
On 17/5/2018 at 8:04 PM, ICDP said:

Unfortunately the 77% extra pixels thing does not = 77% extra image quality.  The PPD on CV1 vs Vive Pro works out about 35% extra.

 

CV1: 456

Vive Pro: 615 (ish)

 

How did you obtain those PPD numbers?

 

The problem with Vive Pro and IL-2 is that all those extra pixels (or extra PPD) are not used effectively. If all the VivePro pixels would be used as effectively as in the Rift, then the VivePro images will be much better. I didn´t try any other games with the VivePro.

Link to post
Share on other sites

Hi Chili, the PPI or PPD numbers are available with a quick google.

 

If you are referring to the excess pixels rendered outside the FoV then that seems to only happen in IL2.  Or we are assuming what is rendered to mirror is exactly what we see in HMD and not just a rough 2D approximation.

 

I tested Elite Dangerous, DCS and Assetto Corsa and saw no FOV issues between screen mirror view and what I saw in the HMD.

 

 

Edited by ICDP
Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...