It will- Its the same as described above. U16P13 and U14Pin11 are the same signal.
Yep …. I went about the mod whilst you sent you suggestions …. 😊😊
mein CBM 8032 (No. 2) mit Commodore High Speed Graphik
-
-
...and the Eltec petGRAFIK is also already laid out, i should go to the prototype stage now by finally ordering some boards
Soso?! Du weißt ja...
Bei der Eltec-Grafik würde ich mich gerne als Alpha/Beta-Tester zur Verfügung stellen. Und parallel dazu mein Original-Eltec-Board versuchen als Laufen zu bringen.
Ich bräuchte nur eine Platine. Und falls zitruskeks sowieso Material, bestellt nehme ich auch gerne einen Materialsatz.
An den Sachkosten beteilige ich mich natürlich.
-
Created a machine code routine for the HSG board to take a bitmap image created using zitruskeks bmp2hsg python code and draw it on the PET screen
-
-
AWESOME!
The memory is really isolated from the CPU, this was never intended. Great work!
-
AWESOME!
The memory is really isolated from the CPU, this was never intended. Great work!
I read a paper that says there can be a way to write directly to memory but don’t think that got implemented.
-
Here is the .D64 image. Load hsgimage and the enter the filename of the image file (do not include the extension .prg or .seq)
The assembly code is hsgimage.asm
Size the images according to the HSG board you have
-
-
Yeah ill publish it tomorrow.
-
But for the moment
-
Writing the bmp display code for the HSG got me thinking (I know, need to get out more) ... what about other animations. I found an apps note for the GPU that draws a moving square which I converted to 6502 and it works rather smoothly (I am having to use delays to slow the drawing down) using machine code. The graphics board can be V. fast if you use delta vectors ...
Drawing a rotating cube (no hidden lines) in basic for example works well but flashing occurs for some lines due to the time it takes to draw the whole image.
I am guessing using look up tables in machine code this would be very smooth - though using sin/cos tables may still be fast enough.
-
When using the EF9366 (instead of EF9365) you get a lower resolution but 2 pages to switch in between.
I did something like that here: RE: CBM 8032: Mein neues Restaurationsobjekt
(there was an error in the program leading to a rest of flickering; I did not post the video of the fixed version...
) -
When using the EF9366 (instead of EF9365) you get a lower resolution but 2 pages to switch in between.
I did something like that here: RE: CBM 8032: Mein neues Restaurationsobjekt
(there was an error in the program leading to a rest of flickering; I did not post the video of the fixed version...
)Yep - mine is using the EF9367 on a 8296-D so 512x512 with very little overall flickering (The 9367 seems to work better on the 8296-D than the 9365) . I could change to EF9366 and use page banking I guess ... will post videos of the rotating square (machine code) and rotating cube (basic)
-
Attached are the videos - the square is slightly slowed and then slowed down by a lot - machine code.
The rotating 3d box is in basic using delta vectors as much as possible. The vertical lines are the last to be draw and show the flickering the most.
-
When using the EF9366 (instead of EF9365) you get a lower resolution but 2 pages to switch in between.
I did something like that here: RE: CBM 8032: Mein neues Restaurationsobjekt
(there was an error in the program leading to a rest of flickering; I did not post the video of the fixed version...
)When switching to my 512x256 on the 4032 the animation is much smoother but slower when paging BUT I cannot seem to clear the screen on page 2... I get vertical dots ... This board uses 4116 ram chips instead of the 4164 on the 512x512.
have you seen that before. Page 1 works fine. Wondering if it is down to RAM chips used ....
-
No I didn't notice any glitches there.
What did you change? I'm confused by your 4116. Which board is this (I did not read this entire thread)?
Shall I test your program on my 8032?
-
Oh. I'm not in my cellar but I just looked at the schematics of my LSG (https://github.com/InsaneDruid/Low-Speed-Graphik).
I did not realize that there are 4116 on it. So I use 4116, too.
-
No I didn't notice any glitches there.
What did you change? I'm confused by your 4116. Which board is this (I did not read this entire thread)?
Shall I test your program on my 8032?
OK - I have 2 HSG boards
(1)HSG 512x256 on my 4032 - 4116 RAM
(2)HSG 515x512 on 8296-d - 4164 RAM (doesn't have the -5V on the board)
the mode (0) works fine on the 512x256 ... its only page 2 that has the vertical dots which is odd as thought if there was a memory issue it would show up in both pages..
My code is - think that's is the one
-
-
Hmmm
Maybe you want to check U30 (LS157). Its the only IC that isn't used at all in 512*512 Mode as its sole job is to create the "MLS3 surrogate"
In 512 mode the 15th address bit of the memory location, generated by the EF Processor. In 512 Mode the processor doesn't outpout it. So i guess that's where the bank switching occurs?
-
My code is - think that's is the one
I tested your C4. I just had to insert a "display(1)" because I did not spend a switch for selecting text/graphic by hand.
There were no glitches in the frames: forum.classic-computing.de/index.php?attachment/203133/
The video above is broken. Here again: VID_20240717_183027344.zip
-
Display More
Hmmm
Maybe you want to check U30 (LS157). Its the only IC that isn't used at all in 512*512 Mode as its sole job is to create the "MLS3 surrogate"
In 512 mode the 15th address bit of the memory location, generated by the EF Processor. In 512 Mode the processor doesn't outpout it. So i guess that's where the bank switching occurs?
Ok will check ... if some of the ram is faulty it would show in page 1 as well is my thought ... the pattern is all over the screen
-
Found the problem …. A 4116 IC was not soldered in right and was half out from the board. Replaced it and now works…. Would you believe it but is was faulty from the start😂
-
Need to make a correction .... the 512x512 HSG board uses MK4516 ram chips which do not need -5v
-
Meanwhile in the basement…
512x512 is a really nice one!
-
Meanwhile in the basement…
512x512 is a really nice one!
Really nice - you used the hsg image code ?
-
I used the python program written by zitruskeks and your *.D64 … a perfect fit! 😍
-
Fantastic .... glad it worked well ... image looks great.
I have modified the code to ignore blanks (0 byte) quicker but speed improvements only slight as it depends on the image
-
Pretty boring really but the HSG board can handle rotating cubes, for example, OK provided the calculation steps do not consume too much CPU time.
I created some code in assembler that takes a pair of points (move to .. lineto), calculates the direction parameter to send to the Command register (the EF926x has a bit format depending on direction and angle), the delta X,Y values to use, and then write them to the EF936x GPU via the respective registers for fast vector based drawing. after each rotation I send a simple clear screen and write the next projection. No flickering so it runs fast. I can slow it down further with a delay routine but speeding it up can only come from pre-calculating all the parameters ahead of time and then just simply send them to the GPU in sequence. May try that.....
I calculated the coordinates in Excel and built a table in the code for almost fastest speed. It is doable to create code to generate the coordinates themselves for each rotation but that takes more time.
-
Oh, really looks smooth!
