Let me know if you wanna try this too

[belzebub]

I added a feature to my mod so you can play lapsilap's world record yourself (same pieces). You can play any replay file again and use an AI to give a hint where the next piece comes.

Download link

Nullpomino Replay Mod [Download]

CHALLENGE

I challenge you to replay the world record of Lapsilap as fast as you can (so it's all finger speed).

Current rankings (top of my head):

1. Paradox 19.95s - 0 Finesse faults
2. Gaviel 39s - Finesse unknown
3. Belzebub 41.23s - 8 Finesse faults

You can share your timings here. I think Paradox has the fastest fingers :o

GETTING STARTED

Basically this is how it works:

To sprint with the same piece sequence/bags as in a certain replay file:

• load the replay file in the replay section (view it once)
• go to start -> line race -> replay-latest-replay

To have an AI to give a hint where to put each piece the same as in the replay file:

• go to options -> AI settings -> AI Type, put it on REPLAY-AI

To load an old replay file of an earlier version of Nullpomino

• first save the replay file in the new style as follows:
  
  • put the replay file in your replay directory
  • go to options -> general options (page 3) - enable "force replay re-save after replay"
  • just view the replay as you do usually until it ends, this generates a new replay file with as name the date and time (as usual)
  • go to options -> general options (page 3) - disable "force replay re-save after replay" (to prevent further copies)
• then just do same as above with the new file

DEMO (without AI hint) and original LAPSI sprint

[Alexsweden]

Wow! - really cool work there! Dont have time to test it now but I would want to test it!

[Sisu]

You read my mind.  

Can you do the same for dig modes?

You can download the best rep files from from netplay.

I think there are too many of these NullpoMino mod threads though. Couldn't they all have gone in the NullpoMino thread?  

[Paradox]

I think there are too many of these NullpoMino mod threads though. Couldn't they all have gone in the NullpoMino thread?  

I agree that there are too many threads, but putting them all on that nullpo thread means:

1. Not as many people will go there if they are searching for it
2. not as easy to directly link to.

[Sisu]

I agree that there are too many threads, but putting them all on that nullpo thread means:

1. Not as many people will go there if they are searching for it
2. not as easy to directly link to.

1. http://harddrop.com/forums/index.php?act=S...ighlite=poo+sfx
2. http://harddrop.com/forums/index.php?s=&am...ost&p=42641

[belzebub]

You read my mind.  

Can you do the same for dig modes?

You can download the best rep files from from netplay.

I think there are too many of these NullpoMino mod threads though. Couldn't they all have gone in the NullpoMino thread?  

Haha great minds think alike -_- I think this is one of the best things I changed to make progression. When reconstructing with perfect finesse I learned things I would never have thought about myself.

Also my brain learns patterns now that I didn't knew before or didn't use because I was used to playing otherwise. By constantly repeating I reprogrammed my brain. Eg for the das to wall rotate drop S and Z. Or the L and J positions on the side.

Or eg I tried to stack flat always. But look at the 3d piece already after T and I, he places the Z on the left, creating a 3 high gap but losing less time by das'ing to the wall then he would by tapping twice and keeping the stack flat.

There are some (for me) complicated parts in the sprint and I feel it's amazing that Lapsi could do them so fast. I now have even more deep respect.

The best part is you can just repeat the sprint and learn. After that I remarked that my own sprints were a lot faster suddenly and that I played with much better finesse overall. Especially for new players this is might be a huge win.

I'll think about the dig modes. Maybe just a map and then throwing the right bags.

I was actually thinking of throwing it in the Nullpo thread. But that thread has already 49 pages. Also this is not really part of my mod, I just wanted to experiment with this and I thought it was worth a new post as it deserves attention. I put plenty of effort in first recreating the sprint and then learning to reproduce it sub50 speed, and I think a lot of people would be interested, so I figured out it deserved it's own post. Imho for a forum that is okay. I guess it's just temporary that there are many topics.

Glad you like it too :-)

[Anonymous]

Wow, replaying someone else's game sounds like a really awesome mod!

How did you replay his game? Did you just have to copy his game a few pieces at a time?

Just an idea, but it'd be sweet if the mod showed the player's (in this case Lapsilap's) next piece placement. So at the beginning of the game it has a kind of ghost piece where his first piece was placed. When you place the first piece, the game displays a ghost piece of where his second piece is placed. Etc. And I guess if you don't follow his piece placement, the ghost piece previews will just stop or something.

This could have a lot of useful applications for teaching other players how to play and teaching techniques. So far, all of the guides on Hard Drop are just "look at this diagram to learn how to do the TKI setup" or "go through this fumen and learn how to downstack this", etc. And they're successful to a degree, but I think new players (and even veteran players) would have a lot to gain from actually playing out these situations (instead of just looking at fumen diagrams), and playing other people's replays and seeing what kinds of things they do when they play.

There are other useful applications too! Like if you play someone's replay, you can see what kind of things they could improve.

Anyways, this is just an idea, but I think it'd be pretty sweet.

[belzebub]

Wow, replaying someone else's game sounds like a really awesome mod!

Thanks, when I play it I feel it's one of the best changes i ever did

How did you replay his game? Did you just have to copy his game a few pieces at a time?

I just replayed the youtube video and copied into the code. I'm now looking into creating an input mode, where you can just place a replay file in a folder and it uses that as input for the random generator, that way it would be so easy to get the same bags from an existing sprint.

Just an idea, but it'd be sweet if the mod showed the player's (in this case Lapsilap's) next piece placement. So at the beginning of the game it has a kind of ghost piece where his first piece was placed. When you place the first piece, the game displays a ghost piece of where his second piece is placed. Etc. And I guess if you don't follow his piece placement, the ghost piece previews will just stop or something.

This could have a lot of useful applications for teaching other players how to play and teaching techniques. So far, all of the guides on Hard Drop are just "look at this diagram to learn how to do the TKI setup" or "go through this fumen and learn how to downstack this", etc. And they're successful to a degree, but I think new players (and even veteran players) would have a lot to gain from actually playing out these situations (instead of just looking at fumen diagrams), and playing other people's replays and seeing what kinds of things they do when they play.

There are other useful applications too! Like if you play someone's replay, you can see what kind of things they could improve.

Anyways, this is just an idea, but I think it'd be pretty sweet.

I think your idea is gold. I learned so much from playing this sprint and I'm not the only one any more. It's good for stacking, seeing opportunities you wouldn't thing of yourself and also finger training and learning finesse, you can make a fault and replay it every time until you get it right every time.

The ghost piece is maybe not so easy to implement (like not in an hour because i don't know how it works now), but i think it's a nice idea, i would make it so that you can put it on or off in the option menu before the game. Would be so easy to learn patterns, opening games, fixes, best downstacking opportunities, etc..

I think actually for every player there are benefits, like learning not to missdrop, learning to use the right finesse, etc. It's like every tetris belt should have a sprint, from very easy to stack to very difficult situations.

It's true that when you learn a replay, every time you would place a piece different you start to reflect on your choices.

I'm currently at 45s (my normal sprint record is 55), with almost perfect finesse. I have Microblizz sprint now too, thanks go out to Gaviel for giving me the piece sequence for that one.

Anyone that wants to try lapsilaps world record bags, just ask me :-)

[joeybeans]

Yeah but I'm one of those people who aren't as interested in sifting through old threads to learn more about that stuff because I'm just not that interested  I like to just casually look through the forums occasionally.

On the other note, I'm interested in trying this, belzebub! I don't know how you did it but can you contact me on skype? My username's mmmjoeybeans

[XaeL]

Yeah but I'm one of those people who aren't as interested in sifting through old threads to learn more about that stuff because I'm just not that interested  I like to just casually look through the forums occasionally.

On the other note, I'm interested in trying this, belzebub! I don't know how you did it but can you contact me on skype? My username's mmmjoeybeans

Wow belzebub ur good at picking up :O

[myndzi]

Replay ghost would be pretty neat. I believe there's some similar code for AI-guided play you may be able to hijack. There's also code for displaying input sequences; you might consider slow-motion with inputs displayed so that you can learn and practice things like DAS preservation easier.

For all that, be warned: Even great players make mistakes or suboptimal placements. Learning their moves too closely can be counterproductive; I think that using it to learn new things is great but trying to play like someone else is going to turn out counterproductive.

[belzebub]

Wow, replaying someone else's game sounds like a really awesome mod!

How did you replay his game? Did you just have to copy his game a few pieces at a time?

Just an idea, but it'd be sweet if the mod showed the player's (in this case Lapsilap's) next piece placement. So at the beginning of the game it has a kind of ghost piece where his first piece was placed. When you place the first piece, the game displays a ghost piece of where his second piece is placed. Etc. And I guess if you don't follow his piece placement, the ghost piece previews will just stop or something.

This could have a lot of useful applications for teaching other players how to play and teaching techniques. So far, all of the guides on Hard Drop are just "look at this diagram to learn how to do the TKI setup" or "go through this fumen and learn how to downstack this", etc. And they're successful to a degree, but I think new players (and even veteran players) would have a lot to gain from actually playing out these situations (instead of just looking at fumen diagrams), and playing other people's replays and seeing what kinds of things they do when they play.

There are other useful applications too! Like if you play someone's replay, you can see what kind of things they could improve.

Anyways, this is just an idea, but I think it'd be pretty sweet.

Replay ghost would be pretty neat. I believe there's some similar code for AI-guided play you may be able to hijack. There's also code for displaying input sequences; you might consider slow-motion with inputs displayed so that you can learn and practice things like DAS preservation easier.

For all that, be warned: Even great players make mistakes or suboptimal placements. Learning their moves too closely can be counterproductive; I think that using it to learn new things is great but trying to play like someone else is going to turn out counterproductive.

I implemented it. This mod was the longest programming from all the changes I did. Skip to 3:30 if you want to see it in action, the first part is how you use it (starts from downloading a replay file from the harddrop website and showing new options, until replaying that file with tool assistence). Sorry it has no sound but my family just can't stfu.

[Anonymous]

Wow! That looks awesome man! Are you going to make it easier to use? I got confused by all the steps you did, but the mod looks good!

[belzebub]

Basically this is how it works:

To sprint with the same piece sequence/bags as in a certain replay file:

• load the replay file in the replay section (view it once)
• go to start -> line race -> replay-latest-replay

To have an AI to give a hint where to put each piece the same as in the replay file:

• go to options -> AI settings -> AI Type, put it on REPLAY-AI

To load an old replay file of an earlier version of Nullpomino

• first save the replay file in the new style as follows:
  
  • put the replay file in your replay directory
  • go to options -> general options (page 3) - enable "force replay re-save after replay"
  • just view the replay as you do usually until it ends, this generates a new replay file with as name the date and time (as usual)
  • go to options -> general options (page 3) - disable "force replay re-save after replay" (to prevent further copies)
• then just do same as above with the new file

Normally, you just get a replay file from someone, put it in your replay directory, view it once and then play it (with AI hint on or off).

It's maybe not the easiest possible, but it's a trade off because it took me already long to implement. You think this is easy enough? The video was everything in one video and i admit it is probably very confusing..

[Anonymous]

Ohh okay, that's a lot easier than the video made it out to be. As for simplicity, it seems pretty simple, but it'd have to be tested out.

I'd be interested in trying it though.

It would also be cool if this became the standard Nullpo version (including your other mods too) so that this feature could be more widespread. c: