Ok, well I decided to experiment. Here is a video showing my modded Xbox 360 controller. It now has an LED that lights when up is pressed:http://n4te.com/misc/inputlag/testing123.avi
I then used my crappy digital camera to capture a video of me pressing up on the SFIV main menu. Here is the video, slowed down to about a frame every 2 seconds:http://n4te.com/misc/inputlag/frames.avi
It looks like the camera caught the LED just as it was lighting up. I think we can assume that the button was pressed at the beginning of that frame. Two frames go by without the game's menu reacting, then by the third frame the menu has reacted.
The controller is of the wireless type. I don't have a wired controller. I have a charger pack so I could plug the controller into the Xbox 360 for this test, but I'm guessing the controller probably still transmits wirelessly? I guess testing it wouldn't hurt to see if the reaction time is better.
I highly doubt the game software does any intense activities that would lag when up is pressed on the main menu. If the Xbox 360 itself or the game are adding any lag, it is probably only a millisecond or two.
At 29.97 fps there is 33.36 ms per frame. A better video recorder would give us higher precision. This means that the delay from my button press to seeing the screen change was between 66 and 99 ms. This is the best possible and worse possible lag that the JVC RS10 can have when accepting component input
, assuming my Casio Exilim EX-S10 actually captures at ~30 fps. I have a composite cable I will test soon for kicks, though I would never actually use that. Unfortunately I don't have an HDMI cable to test with. For those that don't know, the RS10 projector doesn't have a VGA input.
Are there any RS10 settings that might help? 66 to 99 ms isn't very good at all for SFIV considering one frame is 16 ms. That means everything I see is 4 to 6 frames behind.
No wonder links are so hard! I broke the bank on this projector with plans to not upgrade for many (5+) years. I really need it to work better for SFIV!
Edit: Thinking about it a little (I'm a software engineer), games are typically single threaded and handle input in the game loop. This means that the game may delay reacting to input for up to one frame. Assuming the main menu is rendered at 60 fps (like the actual gameplay is), the reaction time may be off by 0 to 16 ms. This improves the RS10's best possible lag time to 50 ms, with the worst case scenario still at 99 ms.