Caution: Non registered users only see threads and messages in the currently selected language, which is determined by their browser. Please create an account and log in to see all content by default. This is a limitation of the forum software.
Also users that are not logged in can not create new threads. This is a, unfortunately needed, counter measure against spam. Please create an account and log in to start new threads.
Point is also that the particular screen I use for that setup will not work with Vsync from my ECS V3s or VIC II sync from my Turbo Chameleon 64 V2s so I highly doubt it would work with AGA MK3s Vsync.
Same message for you: Use AutoResolution, it will have the same number of lines every frame, just like you've programmed for your MK2 unit. No guesswork - if the monitor supports 50Hz, then AutoResolution is *very* likely to work, and it's perfectly V-Synced.
Same message for you: Use AutoResolution, it will have the same number of lines every frame, just like you've programmed for your MK2 unit. No guesswork - if the monitor supports 50Hz, then AutoResolution is *very* likely to work, and it's perfectly V-Synced.
Very interesting feature this AutoResolution. So you can still make your custom resolution profile and then just checkmark this 'Autoresolution' feature and get smooth scrolling? So it will not force you to a specific resolution profile? Because I am very particular in making 1:1 profiles. Making the scandoubled profiles overscan show correctly as it would if the screen was plugged directly to RGB on a 15KHz capable screen. For me I usually end up in the 720x576 range for PAL and my perfect results comes if the screen switches to that specific resolution and 50Hz. Making perfect patterns and everything. With the MK2 I have 720x576 in both lo-res and workbench high res laced, so there is almost no switch delay. Yes, I know the more modern MK3 has solved the switch delay, just like the ECS V3. And yes, I know this is perfect for me and my needs, but other people might prefer other types of profiles, like filling the screen with most games that does not use the overscan area.
If so, would this feature be possible on the ECS V3 and TC64/TC64 V2?
So you can still make your custom resolution profile and then just checkmark this 'Autoresolution' feature and get smooth scrolling?
Please read the documentation: Autoresolution will select whole-number multiples of the input resolution for output, likely violating Vesa standards, but acting similar to the old 1990s-style flicker fixers.
If so, would this feature be possible on the ECS V3 and TC64/TC64 V2?
Indivision ECS V2/V3 cannot generate pixel clocks in sync with the Amiga pixel clock - it's all async there, so no Autoresolution. Similar problem with Chameleon: No direct multiple possible, so output needs to be async.
On my AGA MK2 (which does not support Vsync) I actually managed to finetune it to get totally smooth scrolling with no vsync. Something I thought was almost impossible. The screen I use for that system is a Phillips 17 inch 5:4 LED screen. And it too doesn't support vsync anyway. So the fact I managed to finetune my AGA MK2 to have totally smooth scrolling without vsync for that specific screen is very convenient for me. It scrolls just as butter smooth as if I would have had vsync enabled. I experimented around the 50Hz region until I hit the 'magic spot' for PAL totally smooth scrolling. And even before I hit that spot I had a profile where I had almost smooth scrolling. A little jerk/frameskip now and then.
So I think it should be possible to do the same with the AGA MK3 for a screen that doesn't support Vsync.
Hi again Turrican9 !
I was thinking again about what you said... I think I'm pretty close to what you get.
VSYNC being disabled, and with a vertical freq. of 49.9112, *most of the time* there is no tearing at all (for up to several minutes I think - for instance I really noticed it clearly 2 times when playing the Batman Rises demo but it must have occurred more).
Then once in a while I can see tearing in the form of a *very* slow horizontal line that takes a little less than 10 seconds to go from the bottom to the top of the picture. Then, I have no tearing for a few minutes etc.
This is kinda disturbing because it feels really close to perfection. It makes me wonder if I can finetune the Indi further to get completely rid of it ? If so, which parameters should I try to tweak ?
(for instance, I wanted to try to set a vertical freq. of 49.9113 but didn't manage to because the config tool doesn't seem to memorize precise pixel clock values)
For my particular screen 50.196Hz was the 'magical spot' for perfectly smooth scrolling with no Vsync. But this will vary from screen to screen. I remember I used to experiment around the 50.2xx mark and almost got smooth scrolling. And I seem to remember that going over 50.3xx made it worse. Also going much below 50.2xx made it worse. That's why I never tried much higher or lower than that, until maybe 4 or 5 months ago. I was very surprised when I 'hit the spot' at that particular frequency. I worked from a custom VGA mode I made from the VGA mode menu. Based on 720x576.
You just have to experiment. There is no guarantee you can get 100% smooth scrolling with no Vsync on a particular screen. I remember I read reports from other people managing to get 100% smooth scrolling with the AGA MK2. But using their profiles on my screen didn't give me smooth scrolling. I almost didn't believe it was possible until I hit 'the spot'.
turrican9 : thanks for the advice ! I've been following it and I came up with a temporary interval of values.
With a vertical freq. <= 49.9202 : the tearing induces an upward wave
With a vertical freq. >= 49.9210 : it induces a downward wave
The closer to the interval, the slower the wave... And actually, the closer to the "magic number" I supposedly get, the more annoying the wave is because it takes forever to cross the screen (and it takes a few whole minutes to appear)
Currently testing a value of 49.9207. Fingers crossed !!
Jens : I reach the desired v. freq. by randomly tweaking the clock, F/B porch values etc. It doesn't seem to have any other consequence on the picture. Is that OK to tweak those values this way ?
Is that OK to tweak those values this way ?
Sure - modern monitors don't have physical limitations any more. Times when you had the ability to damage a monitor with funny H/V frequencies and duty cycles are over. That said, don't try that on a CRT.
There's always calculation: One line consists of front porch, sync len, back porch and actual gfx pixels. Add that number up, multiply by pixel period and you have the H frequency.
V frequency can only change in increments of hole lines. Same calculation here: front porch + sync len + back porch + actual number of lines is the total frame time.
Should be fairly easy to make a Libre Office spreadsheet for that.
You could also try UMC: https://umc.sourceforge.net/
Pixelclock can't be chosen freely - the PLLs have limits. Could be that during testing, you already have the altered pixel clock.
Yep, none of the modes I obtained with the calculators would work because of this
But I didn't give up and spent some more time tweaking. I came up with a 49.9208 Hz mode (remember ? at 49.9207 there was an *extremely* slow upward wave and at 92.9210 a verrrrrrrry slow downward one :).
After 20 minutes of Jim Power, still no tearing "wave". Is that my magic number ? I'll wait for a few days before drawing a conclusion !
Yep, none of the modes I obtained with the calculators would work because of this
But I didn't give up and spent some more time tweaking. I came up with a 49.9208 Hz mode (remember ? at 49.9207 there was an *extremely* slow upward wave and at 92.9210 a verrrrrrrry slow downward one :).
After 20 minutes of Jim Power, still no tearing "wave". Is that my magic number ? I'll wait for a few days before drawing a conclusion !
Sounds like you may have found 'the number' for you. Another good test is pinball dreams. After the intro there is some vertical scrolling text. Then at the menu there are horizontal scrolling text you can look at. And if you load up one of levels without starting a game it will scroll vertical up and down.
Then you also have the game 'yo joe' which has some nice horizontal scrolling train at the menu,
Sounds like you may have found 'the number' for you. Another good test is pinball dreams. After the intro there is some vertical scrolling text. Then at the menu there are horizontal scrolling text you can look at. And if you load up one of levels without starting a game it will scroll vertical up and down.
Then you also have the game 'yo joe' which has some nice horizontal scrolling train at the menu,
Hi turrican9 !
Oh no, tearing still occurs... I had let the Amiga run for several dozens of minutes and it hadn't trigger. It occurred several times after 10-15 minutes or so.
(the only thing I changed meanwhile is raise scanlines from 25 to 50% but I highly doubt it makes any difference on the timings - but just in case I put it back to 25%)
Jens : speaking of which, do you think in the future it would be possible to add a 33% darken value ? The bump from 25 to 50 is quite important.
@ Turrican9/Jens : is there any other factor that could impact my search for the perfect timing ? As, a thousandth of Hertz precision seems to be *almost* perfect (as you can see, I can have no tearing for a long period of time, then the "tearing wave" is so slow it takes several minutes to disappear.
Anyway, I'm not ready to re-enable double buffering as I really "feel" I am slightly worse at very fast paced games (and for moving the mouse
do you think in the future it would be possible to add a 33% darken value ? The bump from 25 to 50 is quite important.
So far, the calculation is two-step only (shift, add+shift). With a "two and a half" step approach, we may be able to introduce 37.5% and 62.5% brightness settings for scanlines. Peter knows the details, but Timm would have to add this to the config tool as well. For a more fine-tuned brightness approach, we'd need three multipliers of the FPGA, and I don't know if this amount of resources is still available. pwsoft will have to check.
is there any other factor that could impact my search for the perfect timing ?
The one big imperfection is that it will break anytime the number of lines or timeslots per line on the Amiga side changes.
In other words: What works in one game might already break in the next, hence the auto resolution feature is the only thing that really makes sense. The config tool cannot know what number of lines and what number of chipset time slots per line a game will create, and while the game is running, only a very limited algo can detect the Amiga screen mode and make decisions for the output mode. More complex calculations that may require multiple best-effort approaches are currently out of the question, because we do not have a CPU in the FPGA of the flicker fixer (and I have my doubts that we have space for that).
Jens : I understand. Thank you so much for your replies.
Yep 37.5% and 62.5% would be great alternatives.
Regarding the intermittent tearing I get. It's actually kinda strange. As I told you, it's intermittent, and when it occurs it appears as a slow wave coming from the bottom up to the top (or the contrary).
What is strange is that it seems to be triggered in a "global way". For instance, yesterday, it began to occur around the end of the new Batman Rises demo. I quickly rebooted and started a Pinball game. And the wave was still there and had continued its path. (from the bottom to the top).
I started again those demos/games this morning and didn't see the "wave" but I'm sure it'll happen after a certain amount of time.
Anyway, I'll leave it like that (because it can remain perfect for 15 minutes !) and try some other second had screens every now & then till I find the perfect one
Cheers
So far, the calculation is two-step only (shift, add+shift). With a "two and a half" step approach, we may be able to introduce 37.5% and 62.5% brightness settings for scanlines.
According to the register layout documentation there are 16 programmable steps for the scanline intensity on the mk3 (linear in 6.25% step increments). If only 4 can be selected that is a limitation of the config tool.
The last reply was more than 365 days ago, this thread is most likely obsolete. It is recommended to create a new thread instead.
