Um, I created a FPS counter using RenderStepped on my NVIDIA Shield K1 Tablet, and I got a frame rate of 60 frames per second.
Place used: Generic Mobile Testing - Roblox
You don’t get a frame rate of 60 frames per second. You get an average frame rate of 60 frames per second. Actual timing of individual frames can deviate anywhere between 90 and 30. If you are assuming delta time instead of calculating it, your game is already broken.
If it averages out… it’s not broken?
I don’t follow. If your client is running at 30 FPS and your code thinks it’s running at 60 then your code is going to be moving things at half speed. It isn’t difficult to calculate a delta time. It’s literally subtracting two timestamps. If you’re using RenderStepped then the delta time is even handed to you.
I use RenderStepped and record the last frames time to get deltas. But if someone used .RenderStepped to calculate 1/60 then I don’t think it’d be a problem if they get 60fps[quote=“Sharksie, post:104, topic:22144”]
individual frames can deviate anywhere between 90 and 30
[/quote]
They’d just cancel each other out, right? If one over-compensates the other under-compensates.
Except on mobile where the FPS is 30. I don’t have a shield but I imagine it has better specs than an iphone, which runs at 30.
They also won’t cancel out if your client is just having performance issues. Delta time isn’t optional. If you are assuming that it’s 1/60 seconds then you are writing objectively bad code.
Wow neat. I still need to store it manually, since I do some ugly math in certain areas which actually consume quite a bit of time.
Yes, absolutely. On VR devices we’ll basically run locked to whatever framerate the headset expects - 90 Hz for both Oculus Rift and Vive. We’ll keep the 60 Hz limiter if you’re not in VR.
We currently run the game at 75 Hz in VR with DK2. I have not really seen games break due to that - as noted above, RenderStepped is already variable framerate.
Depending on your phone and the place we may run at 60 even on mobile, although of course that depends.
VR on mobile will pretty much require us to run at 60 Hz - early experiments w/Cardboard and Gear suggest that while this is not possible for all mobile games, it’s possible for some of them (despite rendering the scene twice - although we had to implement some key optimizations for that, and we’d have to work on that a lot more to get more content at 60 Hz on mobile in stereo).
At my university we have multiple dev kits of both Oculus and Vive, but so far I have not yet thought of a reason to actually test ROBLOX with them considering making them work with properly-supported engines is a pain already. I absolutely love how this emerging technology is being implemented within ROBLOX, although when you compare the potential target audience of ROBLOX players with VR and the limited publishing options I’m afraid ROBLOX and VR won’t add any value to one another any time soon.
Cardboard support might be the best bet here as it is accessible to a large portion of ROBLOX Mobile players, but for myself the ROBLOX Mobile app is still having too much issues to consider using it over the large number of other games available on the over-saturated app stores…
My personal opinion is that Oculus is a good place to test things in development, while Cardboard/Gear/etc. are mainly the end product. They’re mostly the same concept on the highest level, like controls. Some things are obviously different; Cardboard has bad sensors and no real position tracking, Oculus has a cord and is usually used sitting down, and so on.
“Properly-supported” is a weird term It’s mostly the engine developers (I’m assuming you’re talking about Unity and Unreal) who are developing VR support, not the platform vendors.
On desktop, barring the pricing concerns - which I’m hoping will eventually get resolved by devices getting cheaper - there actually seems to be a great synergy. Consider:
- A lot of games on ROBLOX use default controls. This means we can perfect the built-in controls in VR mode and games just use them.
- ROBLOX is heavier on 3D UI than a lot of traditional games. We’ll push for this even further. This is important for VR.
- VR will have a distinct lack of content. If you’re in the target age segment for ROBLOX, the variety we have is great.
- We have much more aggressive dynamic performance scalers than most games. A lot of VR experiences will require top notch hardware - like I have GTX 960 and I have significant performance issues in a lot of VR demos targeted for Oculus/Vive. ROBLOX has a much lower GPU baseline requirement plus we dynamically scale.
The result of this is that many games on ROBLOX will be playable in VR - unless they use tons of 2D UI. This is what we’re shooting for and based on current experience - we are in VERY early stages of adapting camera/character controls for VR - I’d say it’s realistic.
On the side of ROBLOX players - not developers - frankly, playing in VR is exciting. Turns out you don’t need to have a photorealistic experience for it to feel great and immersive.
Case in point - Deathrun works okay as is. There are some issues with in-game UI (hard to read right now) and there are some camera issues that are relatively widespread in other games that we’re addressing internally - I’ll post separately when that ships. But it works. I’m expecting that after we ship that you’d need UI tweaks on your side and some performance work on our side for it to work well. Now extrapolate to 100 games like this.
On mobile, the pricing concerns are not there mostly, but the performance is harder to reach, and controls are much more challenging. We do not know how well our current plans will work there - we’ll see.
CAN I PLAY ROBLOX ON MY MCVIRTUAL REALITY?
I’ll reluctantly admit that if the box has a QR code with a viewer profile we will support it.
Do you think a 1st gen Nexus 7 is up to the task of doing VR? I recently got one for the right price.
… how did you plan on strapping that to your head?
Is it too cliche to answer “duct tape and cardboard?”
If I didn’t feel like being creative I could buy one of these: http://www.yay3d.com
My question was mostly about the device itself instead of the attachment. I am an engineer. If the device will do VR rendering, I will make it fit on my head.
So, any word on 1st gen nexus 7 doing VR? I want to know before I try to fix the tablet.
It’s hard to say. Our mobile VR efforts are very early right now - we only tested on a few phones and are still optimizing.
Important factors are:
- Performance. CPU and GPU are both important. Nexus 7 seems like it could work maybe, although it’d be better with a better phone.
- Gyro latency/accuracy. No idea what it’s like on Nexus 7 but this is critical.
- Graphics drivers. We mostly worked with ARM Mali so far, Nexus 7 has Qualcomm Adreno which may require tuning on our side for some timing stuff.
Also note that mobile VR support will take us - I’m guessing - a couple of months to ship.