A simple & short TAS. The goal is to complete Training Area from the title screen.
This is a 4-frame improvement over my TAS 3 years ago. 2 frames come from this new Air Dash technique I found during my X4 Zero TAS. There's another 4 frames difference due to newer emulator version, but that doesn't count as an improvement.
Hi HappyLee.
Regarding the topic https://tasvideos.org/Forum/Topics/26020 , you mentioned that there was a one-frame input delay of some kind. According to the submission notes, you used BizHawk 2.10. Did the one-frame input delay have any effect on the TAS?
Regarding the topic https://tasvideos.org/Forum/Topics/26020 , you mentioned that there was a one-frame input delay of some kind. According to the submission notes, you used BizHawk 2.10. Did the one-frame input delay have any effect on the TAS?
Good question. I actually did this TAS with BizHawk 2.8 Octoshock (no delay), and then resynced it with 2.10. The actual game play should be the same. Difference is with lag frames.
Unfortunately my issue in that topic hasn't been solved, so I have no choice but to do X5 TAS this way. Hopefully there'll be better ways in the future, because resyncing a TAS takes extra efforts.
Recent projects: SMB warpless TAS (2018), SMB warpless walkathon (2019), SMB something never done before (2019), Extra Mario Bros. (best ending) (2020).
Regarding the topic https://tasvideos.org/Forum/Topics/26020 , you mentioned that there was a one-frame input delay of some kind. According to the submission notes, you used BizHawk 2.10. Did the one-frame input delay have any effect on the TAS?
I actually did this TAS with BizHawk 2.8 Octoshock (no delay), and then resynced it with 2.10.
Can you upload your BizHawk 2.8 Octoshock (no delay) TAS, for comparison purposes?
I would like to understand what you mean by "no delay", because I tested Mega Man X5 on both cores in 2.8, and I was unable to find a difference so far.
I would like to understand what you mean by "no delay", because I tested Mega Man X5 on both cores in 2.8, and I was unable to find a difference so far.
Well, the delay is in how the emulator responds to an input. For example, in Octoshock 2.8, if you press X to jump, it will only take 2 frames for the animation to appear on screen, while in later versions or Nymashock, it will take 3 frames.
Can you upload your BizHawk 2.8 Octoshock (no delay) TAS, for comparison purposes?
Sorry, I don't share my test version of this TAS. Besides, you can't see the difference from the movie file. The difference is with how the emulator responds to inputs, not only from the screen image, but also for memory values as well.
Recent projects: SMB warpless TAS (2018), SMB warpless walkathon (2019), SMB something never done before (2019), Extra Mario Bros. (best ending) (2020).
Well, the delay is in how the emulator responds to an input. For example, in Octoshock 2.8, if you press X to jump, it will only take 2 frames for the animation to appear on screen, while in later versions or Nymashock, it will take 3 frames.
I tested again as follows: Load "Mega Man X5.cue" (not bin). From Reboot Core, enable frame count and input display, go straight to the intro stage (starting game and clearing all dialogs to get player control), don't move until the emulator says on screen frame "4676". Press "X" and frame advance at the same time, so the display now says "4677 X". Hold "X" and press frame advance until the player first goes into the jumping animation.
I tested both cores, both player characters X and Zero. In all cases, I found that the player goes into the jumping sprite when the emulator says "4680 X". That is, I found it takes 3 frames for the animation to appear in all cases I tested, being unable to find any difference between the cores or the characters. For reference, the emulator's About screen says that the version of BizHawk I am using is "Version 2.8 (x64) (GIT HEAD#e731e0f32)" and "Version 2.8 (x64) February 19, 2022".
Not saying that emulator differences in input don't exist, just saying that I am unable to replicate them.
HappyLee wrote:
Unfortunately my issue in that topic hasn't been solved, so I have no choice but to do X5 TAS this way.
You may wish to consider the likelihood, that in fact BizHawk 2.10 is the emulator with correct emulation of input latency, and previous versions of BizHawk are inaccurate.
General consensus is that later emulator versions means the emulator is more accurate, so if you are saying that BizHawk 2.10 is less accurate, you will need to convince the developers in that topic that BizHawk 2.10 is inaccurate. Otherwise they won't believe you.
For reference, the emulator's About screen says that the version of BizHawk I am using is "Version 2.8 (x64) (GIT HEAD#e731e0f32)" and "Version 2.8 (x64) February 19, 2022".
My version is exactly the same. I'm using ".cue" as well. It's strange that we get different results. Could there be some other settings that may influence this?
I honestly don't know if newer versions are indeed more accurate with everything, but having one extra frame delay in X5 than X4 is pretty weird to me. Since I've also tested PSXjin (no frame delay), the game code in those two games shouldn't be that different.
I just hope that someone can figure out why it happens, or come up a easy solution to fix it. Otherwise TASing X5 would be extra hard for me, since I do manual inputs.
Recent projects: SMB warpless TAS (2018), SMB warpless walkathon (2019), SMB something never done before (2019), Extra Mario Bros. (best ending) (2020).
I tested both cores, both player characters X and Zero. In all cases, I found that the player goes into the jumping sprite when the emulator says "4680 X". That is, I found it takes 3 frames for the animation to appear in all cases I tested, being unable to find any difference between the cores or the characters.
For me, it takes 4 frames (frame advance 4 times) to see jumping on screen in X5. 3 frames are fine (same as X4), 4 frames mean an extra frame delay.
Recent projects: SMB warpless TAS (2018), SMB warpless walkathon (2019), SMB something never done before (2019), Extra Mario Bros. (best ending) (2020).
I tested both cores, both player characters X and Zero. In all cases, I found that the player goes into the jumping sprite when the emulator says "4680 X". That is, I found it takes 3 frames for the animation to appear in all cases I tested, being unable to find any difference between the cores or the characters. For reference, the emulator's About screen says that the version of BizHawk I am using is "Version 2.8 (x64) (GIT HEAD#e731e0f32)" and "Version 2.8 (x64) February 19, 2022".
Yep, if you test against an older version (I tested against BizHawk 2.3), you will notice it immediately.
