Hybrid Scanlines

Viewing 15 posts - 16 through 30 (of 98 total)
  • Author
    Posts
  • #19519
    marqs
    Participant

    @paulb_nl does the implementation set per-pixel mask intensity based on the original value alone, or does it need data from neighbor pixels? As for line-multiple specific adjustment feature, I haven’t had yet time to implement that but basically it’d just move selection logic (which sets V_SCANLINEID) to soft-CPU side where it can set more dynamically. You may want to check a pending pull request from borti4938 that adds another generation method and overall changes the code a bit.

    #19520
    Morpheus_79
    Participant

    There’s a firmware test build for bortis scanline method available on his own github repository:

    https://github.com/borti4938/ossc/tree/dev_lcdbl/fw_build

    #19526
    James-F
    Participant

    From what I see the 0.79 scanlines are already hybrid, and the “scaline str. %” parameter controls from what IRE the scanlines will start to blend gradually.
    ie. A setting of 25% will have not effect on IRE 25 and below so they will have 100% scalines, but will gradually brighten the black lines from 25 to 100 IRE.

    Kind of looks like a simple linear equation where the slope is movable horizontally, as such:
    https://s10.postimg.org/fzov7td5l/scanline_slope.png

    The missing parameter is how quickly (slope) the filling of black lines occurs from the selected strength % parameter.
    In that sense borti4938’s and paulb_nl’s final result is quite similar but differ at what point and how quickly the blending occurs.

    #19530
    paulb_nl
    Participant

    @marqs Yea basically its just subtracting the original value from the scanline strength so that high values have lower scanline strength. Thanks for the heads up about the pull request. I will check it out.

    Here is the link to the test firmware for people that want to try it out. I left it running for 4 hours and there were no issues so it should be good. https://drive.google.com/file/d/1pNzelKMLYHA6E0KJgsj0f0Y34Bw9uQrO/view?usp=sharing

    #19531
    James-F
    Participant

    Works great!

    Playing with the contrast and strength setting and indeed the resulting gamma curve with the “Low” 87% setting is much closer to what a CRT looks than the standard Off setting at any strength,,, it’s all about the resulting gamma curve that looks nice or not.
    It seems the High setting is much too much and not very useful below 100% strength and Low is also slightly too much when below 75%.
    I think the sweet spot between Off and Low is missing.

    paulb_nl can you try adding a setting between Off and Low? Maybe change Low to Medium and add another lower Low.
    And maybe move the Contrast setting below Strength as they are work together to achieve the result.

    My reference is a Sony Trinitron 29″ CRT with the lights off which looks beautiful.
    I’m trying to achieve the same look on the OSSC with the new scanline settings.

    #19532
    borti4938
    Participant

    Hey @paulb_nl
    nice that you have implemented ‘hybrid scanlines’. 🙂

    Would you mind to share the code on a new GitHub branch at this point of development? If you have the feeling that just fine-tuning has to be made, I would love to include that into the pull request; and make it user selectable such that each user can select the flavour he likes. I’ve added a menu point anyway to compare directly with the ‘old’ method.
    Also, would you mind if I use your implementation in my N64 project?

    BR

    #19533
    paulb_nl
    Participant

    Hi @borti4938 Sure you can use it! Its not a very complex implementation anyway 😀 I have named it Scanline contrast for now. Here is the commit: https://github.com/paulb-nl/ossc/commit/990bc1563eef0a5650dabcc5633d3aeb562abc38

    James-F wants a lower setting so I was thinking to add 50% Low alongside 62% Medium and 87% High.

    #19534
    borti4938
    Participant

    Cool – thank you 😉 That’s simple, indeed.
    Will merge it soon at the begin of next week.

    I’m thinking of adding it into the methods as Hybrid (low), Hybrid (med) and Hybrid (high) next to Multiplication and Subtraction in the “method” selection.

    #19537
    Retrorunner
    Participant

    I have tested bortis scanlines and it looks better. Colors are much closer to the original value:

    bortis scanlines

    bortis scanlines

    #19543
    James-F
    Participant

    James-F wants a lower setting so I was thinking to add 50% Low alongside 62% Medium and 87% High.

    Thanks, yes a lower setting than the current low is needed.
    On the current Low setting the white color starts to have scanlines at 75%, and the resulting gamma is too high.
    While Off (stock), white starts so have scanlines immediately, but the image appears to be flat (gamma too low).

    A great way to look at the resulting gamma curve with scanlines is the “Grey Ramp” on the 240p suite.
    Put Strength on 50% and flick between the Contrast settings.
    grey ramp

    #19546
    paulb_nl
    Participant

    @borti4938 Sure that should be fine. Thanks


    @James-F
    That is indeed a great way to see the differences. The lower Low setting at 50% scanline strength now shows scanlines up until the brightest bar so that should be the sweet spot you are looking for.

    #19547
    borti4938
    Participant

    I made some notes and from principle it’s not a problem to make the ‘contrast’ value for hybrid scaliness freely available from 6.25% up to 100%. There are enough free multiplier units left in the FPGA.
    Although I feel that it makes settings a bit too complicate for the casual user: “‘scanline strength’ and ‘contrast for hybrid scanlines’…??? Which one should I vary first???”

    What do you think? @paulb_nl, @marqs

    #19549
    paulb_nl
    Participant

    Indeed I think it would make it too complicated. Also if you go too low with the contrast setting then the difference is not much noticable between Hybrid off and on. 50% is probably the best lowest setting.

    #19554
    borti4938
    Participant

    OK, my intuition, too. Then I will left it as is with the three presets and will left one bit as reserve.

    #19562
    Retrorunner
    Participant

    Tested and I think we are on a good way but the main problem remains as with the standard scanlines (via substruction): the color level on dark surfaces is way to crushed and looks false (check the wall on the left).
    I think the combination of bortis (scanlines via multiplication ) and hybrid scanlines would be near perfect – if possible.

    To zoom in: https://img1.picload.org/image/daagolri/scanlines.jpg

Viewing 15 posts - 16 through 30 (of 98 total)
  • You must be logged in to reply to this topic.