Printable View
Would it be possible to add pasting quick backdrops from an active/active picture?
Specially because of motif mode, it seems a lot more efficient than having multiple backdrops representing repeating patterns...
Or at least I want to know if it can be done as an extension with the SDK...
Hi Yves, thank you for worrying, I fixed most and did a work around for the animation "bug"? I will explain here and be of further assistance... (I faced 3 problems)Quote:
Originally Posted by Yves
Problem 1: Focus box was not working
Solution: I discovered this was due to a pause and unpause with keyboard key P, I had done on the global event editor
the the game paused when a person would type P on the username and password screen.
Problem 2: Global Escape to end application would not work in that edit box screen either
Solution: This was because the game was paused, lol - Fixed after deleting the global pause with P (same fix as problem 1)
3- the game won´t even get past a screen even though the event has always been there before...
Solution: I still do not know as I have not made further tests... I will do it right now and see again....
I had a screen (frame) which only had an active in it with a one frame only of animation without looping set for an animation speed of 10
So before the game would auto jump frame after that "animation" was over... it did not yesterday, after update... so this may be the only thing you may want to look into, but let me make sure that is really the case and then I will see if not I will send you an mfa! because why waste your precious time if I could fix and find out.. see you soon...
The code that handles pasted backdrop objects doesn't handle quick backdrops, you can't do this in an extension. I agree that pasting quick backdrops would be more optimized for repeated backdrops, but significant changes are required for this.Quote:
Originally Posted by NaitorStudios
OK thanks.Quote:
Originally Posted by Aloan
Problem 1: I tried what you said and it works the same in previous builds. Just to confirm.
Problem 3: I tested this too, via "Animation Stopped is over / Next frame", and it works, so yes I would appreciate an example if you can find what happens and think it's a bug in the new build.
it might have been due to the little devil Pause Unpause with P that conflicted with my other 2 previous problems.... So in the end there is really nothing for you to worry about Yves! it has nothing to do with the program itself. Thank you! Fusion ftw!
BUG?
Cannot retireve the text of the String object if it created by event.
Attachment 29777
Hmm, don't know if this is intended or not, but both "Paste into Background" and "Add Backdrop" from the Active object is doing exactly the same...
Both are using the (backdrop) object limit, both can be destroyed during runtime...
The "Paste into Background" documentation says:
- objects pasted by this function only exist in the part of the screen that is visible at the time it's pasted. If the object goes out of the screen, for example in a scrolling application, it will be definitely destroyed.
But it doesn't seem to happen... ?
So what's exactly the difference? o.o
I thought Paste into Background was supposed to not be affected by object limit, and perhaps perform better...
(would be nice if we could paste cropped parts like Active Picture does too, it feels like Active Picture is slower in hardware acceleration mode)
Edit: It seems like "Paste into Background" only destroys pasted stuff that goes outside the view area in Standard mode
Yes "Paste into background" is an old 'standard graphic mode" thing, in DirectX mode it works like Add Backdrop.
It's seems it's part of the weird "feature" that prevents string objects created in frame editor from being destroyed. I *think* I can fix this part without causing issues, this will be done in the next beta.Quote:
Originally Posted by ASD
Yves, thanks for the info about "Paste into Background"!
I noticed a problem while doing some tests (not related to the previous thing)...
I have a 1080p game, but when I put it fullscreen on a 4K monitor, the GPU usage increases and the performance drops by a lot... (Idk if Windows scaling also affects it)
So I guess Fusion renders like 4 times the pixels instead of just stretching the image (?)
Is there any way to optimize this?
Are you in DirectX 11 mode? In this mode Fusion doesn't change the resolution (I couldn't find a way to do it yet with the current structure of the display in Fusion apps...), it stretches the display. But stretching a HD image to 4K takes time even for a GPU. This is the main thing we still have to improve for DX11.
In DirectX 9 mode, it changes the screen resolution, so this would be probably a bit faster.
Okay, so when going fullscreen in DX9 it drops 8 fps, in DX11 it drops to almost half the fps...
One thing tho, Change Resolution mode isn't exclusive fullscreen, right?
Because it messes with other window's positions and size when switching the resolution, this is the main reason I avoid it...
Would Exclusive Fullscreen be possible?
It's exclusive full screen in DirectX 9 mode. It's not exclusive full screen in DirectX 11 mode yet.Quote:
Change Resolution mode isn't exclusive fullscreen, right?
Oh I see, I remember Standard mode messing with other windows in change resolution mode, this seems to not happen in DX9 which is great!
Although with change resolution mode on, DX11 gets a bit glitchy, the frame gets off-centered (bigger border on the right), but it's nice to know DX9 already got Exclusive Fullscreen.
Hope to see it in DX11 in the near future if possible, or at least some optimizations to get better performance in borderless fullscreen mode.
Hmm, what's your app window size, frame window size and current screen resolution?Quote:
Originally Posted by NaitorStudios
320x200 with Change Resolution Mode enabled, works fine on DX9, gets weird in DX11
OK I'll take a look.Quote:
Originally Posted by NaitorStudios
For info the build 293.1 is no longer a beta version.
Hey @Yves , Can you add options to the list object to make it a global object? I keep my data there and want one instance for the entire game. The second thing is a bug. If I create an object that doesn't exist in the current frame and exists in another frame with create action object it shows up, but if I do it with global events then it doesn't work. I wanted this object to be automatically created in every frame so I put it in global events.
Global events (Did not work)
* Start of Frame
New Objects : Create BaseData at (-32,-32) layer 1
Each game frame(It works)
* Start of Frame
New Objects : Create BaseData at (-32,-32) layer 1
Note that BaseData exist only in the base frame
This is a known issue and we cannot change it easily: Create Object works in global events only if the object exists in the frame (you can unselect its Create At Start option if you don't want it to be created when the frame starts).Quote:
Originally Posted by aa2953b4
What about list object? I need to keep data from previous frames one global instance of list object as in the case of array or active would be an ideal solution here.
you could save and load list content if no option for global is available?
I create a clever framework for quickly developing games of a given type. Therefore, I try to avoid unnecessarily copying the same code in each frame. Loading the list in each frame is pointless, especially since other objects are used to initialize/handling the list and are redundant in the next frames.
I have no longer false positive was finally able to create build on this version that was not deleted/detected by avast. I picked unpacked exe option. Didn't try it further (packed). Also it's very appealing to have separated files and exe with lower size. Can't wait for info on Fusion 3, all the best.
A major system in my game broke with this new update (also broken in all old versions of the game, so I'm certain it's the update.) Can I roll back the update with the steam version of CF2.5?
@dustingunn Steam don't allow rollbacks, either way would be better if you help us find and fix this for the next updates... What did broke? Could you isolate this?
I've isolated it as much as I can here. There's a big red comment where I've isolated the code that's not firing even though all conditions are met. Even previous, simpler versions of this in old MFAs are dysfunctional in the same way. I notice when making this isolated version, it works for the first enemy placed (in the middle) which might mean something about fastloops/foreach loops has changed and it's not going past the first instance. However, that can't be the case because I used debug to test for the variable that wasn't working and it turns out to be set correctly (GCollideType is being set to 4, but "If GCollideType > 0" isn't triggering.)
Attachment 29802
Also, Steam does support rolling back. GameMaker Studio 2 features full rollback options using the beta build select.
If you pull the lines you refer to, out of the group activation it works fine. Your bug has more to do with the set up of groups, than with the code itself. Attachment 29804
That breaks the collisions. The system tests a large rectangle (based on how far the bullet is moving,) and if there's a hit, it goes back and does a per-pixel test. You moved the code out of the fast loop so it's just testing once instead of per pixel (which is why the bullets clip so far in your update)
Oh I see, I just know that I’ve had problems with using groups in the past, I thought that’s what might be breaking it.
I found a simple bug.
If you have in your frame the "Fade In" option and you have an Array Object then "Start of the Frame - Set a global value" with the array, the game crash.
The problem solves if you delete the "Fade in"
Attachment 29805
I can confirm @dustingunn did broke with the update... Not sure why tho...
But isn't that overly complex for what it need to do anyway? Over 100 events for pixel and bounding box collision seems like too much...
I'm not sure. It used to be a lot smaller but wouldn't work if multiple bullets hit an enemy in the same frame, so I added the nested foreach loops. It's really hard to tell what will and will not work with CF2.5's object picking.Quote:
Originally Posted by NaitorStudios
This is a known issue, and that's why the "Create before frame fade-in transition" option exists. Select this option for the Array object. IIRC by default extension objects are not created before the transition starts, so if you use their expressions in a Start of Frame event you get a crash.Quote:
Originally Posted by Zinume
It's due to this optimization in the new version:Quote:
Originally Posted by dustingunn
"- Windows runtime: performance issue when a fast loop has several Start Loop actions."
I don't understand what's wrong yet, the optimization is quite simple, either there is something I didn't understand in the original code, or the original code was bugged and the optimization fixed it and broke your events, I'm on it.
So I built non packed exe, Avast does not longer detect it as false positive. All works ok.
Antivirus engines will recognise Fusions runtime exe and packing tools as potential threats and flag as a virus.
This is totally normal behaviour from any AV engine. The only reason they don't flag after updates is because a new build has been relaeased into the wild which they don't yet recognise.
Some will false positive still but you should really learn how to use your AV software. Leaving it unconfigured isn't always the best solution and can really hinder performance.
You have to create exemptions for the Fusion applications and folders which are used to build your software so the AV knows that these files are OK and not to clean/quarentine them.
The same thing happens with any Dev environment where the AV has not been correctly configured to ignore some known processes.
Granted that this could potentially allow unwanted threats in. But unless you are going to places and downloading things you really shouldn't be, then the threat is minimal.
It still surprises me that Clickteam have not used or have some sort of software signing as part of their build process before releasing.
This can help mitigate the distribution of unlicensed versions and an extra layer of trust that should really start to be used now.
https://www.ghacks.net/2018/04/16/ho...ms-in-windows/
Good work thanks Yves.
Regarding this update:
- Windows runtime: List object now supports up to 65534 lines (it was limited to 32767 before).
is there a complication to make lists global?
I can't get it updated. For some reason it stopped working in my computer when I changed CPU and I had to reinstall from scratch. Now the updater says that there is an updated and it begins to download the update - then it never finishes. It continues to restart the download of the patch for ever ("downloading update patch" dialog).
What to do?
Isn't there manual download for the patches nowdays?
No, it's not the reason, for a few builds they were flagging very shortly after the update. I think it's because when they analyze an EXE, their heuristic engine take various things into account to flag it as "possible malware", including the Win32 API functions it calls, signatures and caracteristics of various parts of the EXE, etc. and for some reason the previous builds were close to the border between "possibly safe" and "possibly malware". In the build 293.1 we've worked a lot to remove several function calls that could be considered as suspicious, like using Windows hooks that have been replaced by potentially less suspicious routines, etc. Also we've changed some important parts of the code to change the signatures. I think this makes the runtime look less suspicious to AV's.Quote:
Originally Posted by DJFuego
Our main program is signed, as well as the runtime used by the editor. But obviously we cannot sign the stand-alone runtime that is used to make stand-alone applications. Firstly as soon as we build an application the EXE of the application gets modified, making the signature sinvalid, and secondly it's the user's responsibility to sign their own applications.Quote:
Originally Posted by DJFuego