Skip to main content

May

Thursday 1st​

  • Experimenting with ways to indicate multiple contacts aggregated together into one
  • Setting up scene for zooming and scaling rings
  • Experimenting with distance invariant scaling
    • Scaling rings with distance
    • Placing text on ring/line intersections
    • Scaling text with distance for constant size
    • "Logarithmic rounding" of distance values
  • Better string caching

Friday 2nd​

  • Experimenting with another way to draw rings, smoothly fading in new distance bands as you zoom out
  • Drawing line segments over arc when view angle is straight on, this prevents arcs disappearing when viewed from acute angle

Tuesday 6th​

  • Added thickening to arcs when viewed at straight on view. Making it more obvious as the view approaches straight on.
  • Experimenting with icons for groups of objects
    • Dots
      • Stacked vertically
        • Show count once max stack size is reached
      • Dice
    • Just numbers
    • Final version:
      • Stacked dots, with number count show for any stack count (i.e. more than 1)
  • Showing different outline on hot and hold contacts
    • hot = triangle corners
    • cold = lines
  • Started pulling out common drawing methods to DrawEx class

Wednesday 7th​

  • Adding straight velocity lines as well as projected arc lines
    • This looks very cluttered and is hard to read
  • Changing line thickness for velocity lines (noots instead of pixels)
  • Extracting some more DrawEx methods
    • Ring of outward facing lines
  • Split indicator drawing out to a separate class
    • Rewriting indicator drawing to handle groups icons

Thursday 8th​

  • Using torus segment instead of arc for drop arcs, this fixes them becoming thin when viewed end on
  • Aligned cold indicators with direction of travel
  • Updating polyline shader:
    • Pass through distance and time as vertex data
    • Threshold in time (smooth cut-off as time passes)
    • Dashes in distance
    • Also dashes in time (acts as an indicator of speed)
  • Updating polyline generation to generate mesh bounds for culling
  • Added constant screen size thickness to polyline, adapting width in vertex shader to prevent it becoming too thin

Friday 9th​

  • Rewriting polyline generator to fix some winding issues
  • Updating tactical view to use pixel thickness
  • Moved polyline points data generation into a job
    • Fix generation with deferred arrays
  • Added support for tinting/blinking lines

Monday 12th​

  • Storing time in polyline vertices
  • Testing dashing of velocity lines
  • Added caching for text elements in velocity lines
  • Handling corner case of arc projection when passing over poles
    • Drawing dropline instead of droparc
    • Tweaking position a tiny amount by velocity, so it's not over the pole
  • Fixed generation of invalid bounding box for polylines with no points
  • Added "direct" lines (from inner sphere to position)
  • Added "inner arc" (arc from XZ plane to projected position on inner sphere)
  • Added sphere showing ship orientation
  • Text scaling for velocity lines
  • Measuring outer radius of rings and clamping velocity lines to end at that point
  • Adapting camera near/far plane to radius of tactical view allowing for infinite zoom (and inevitable Z-buffer precision issues in the future)

Tuesday 13th​

  • Adding ship attitude indicator
  • Building entity system sim for tactical view
    • Adding ship entity
    • Adding contact entities
  • Creating components to represent contact data
  • Investigating representing track numbers using other bases

Wednesday 14th​

  • Generating some ISR ideas [[Design/ISR/Ideas]]
  • Experimenting with SIMD in basic Myriad.Dynamics integrator
    • There's no 256 bit fmadd_ps instruction :(
  • Copying contact data out from ECS
  • Improved ship attitude indicator
  • Fixed string cache returning incorrect strings when large
  • More precise timing info for end of path
  • Better time formatting for end of path
  • Experimenting with kalman filtering to sensors
  • Experimenting with particle filters

Thursday 15th​

  • More particle filter experiments
    • Using particle filter could work for integrating sensor data, but would be complicated to tune. Probably better to fake it.
  • Added lots of helper functions over spans for statistics

Friday 16th​

  • Experimenting with noise generators for sensor simulation
    • Pink noise - Random walk with smoothing, strong low frequency components
    • White noise - pure random gaussian, thermal noise
    • Shot noise - White noise correlated with sqrt(signal).
    • Quantisation noise - Round to nearest N, simulating limited sensor resolution
    • Dropout - Set readings to zero
    • Impulse Noise - generate occasional spikes in noise which last for a short duration.
    • Sensor modelling
      • e.g. instead of adding noise to XYZ position, convert into bearing/azimuth/range from sensor, add noise, then convert back to position
  • Building a system to fake sensor data for a radar dish
    • Sensors can generate a probability distribution from a reading
    • Propagating that into the future can use ground truth for position, and widen the stddev a bit. No need to do true extrapolation.
    • Combine multiple readings somehow
      • True sensor fusion?

Thursday 22nd​

  • Back from holiday!
  • Experimenting with new/simplified sensor model
    • Core values tracked:
      • Range Deviation
      • Direction Deviation
      • Position Deviation
    • Sensors can improve any of the values
    • Values decay over time

Friday 23rd​

  • Drawing gizmos for new sensor model
  • Investigating library for UI animations
  • Added exponential growth in measured deviations as time passes
    • Sensor systems can improve these values
    • Some "sensors" can simply be cross referencing values
      • Created functions to calculate core values from other core values
      • Applying this to all values every frame and taking the best value, emulating "sensor fusion"

Monday 24th​

  • Added a function for integrating new readings (e.g. from radar)
    • Position sense (RADAR)
    • Direction sense (IRST)
  • Added assembly definitions for sensor and DRADIS prototypes
  • Improved zooming acceleration/deceleration curves in close view (looking at a ship)

Tuesday 25th​

  • Improved camera pan/rotation (made them less sloppy with appropriate damping values)
  • Normalising mouse inputs by screen size, to make panning/rotation have a sensible speed at different resolutions
  • Building out ECS data model for visual sensor tracks (i.e. the data needed to drive the tactical display)
  • Building systems to fill in visual model from sim model

Wednesday 26th​

  • Removed support for grouped contacts (one icon representing more than one entity), this removes a lot of complexity
    • Simpler one-to-one data model between visuals, tracks and entities
    • No need to group and ungroup based on camera view
    • No need for UI to force splitting/grouping
  • Finished systems for calculating tactical view data
  • Tweaked appearance of icon (shrinking them slightly)
  • Added changing of direct line (dash and alpha) based on if the contact is currently locked
  • Changed confidence from an enum to a factor from zero to one
  • Using blinking less in the tactical view, too distracting
  • Experimented with dashes for velocity lines
    • Set offset to Time.Time % (DashLength * 2) to "fix" the dashes in place, as if the object is moving along a fixed line
    • Set offset to zero to have the dashes move with the object (i.e. first dash always starts at object)
    • Best option: Set offset to -Time.time * N to have dashes emerge from object and move along track, in direction of movement
      • This gives an indication of speed, even when sim time is frozen

Thursday 27th​

  • Added instanced symbol rendering to tactical view scene
  • Investigating how to make symbols clickable
    • Create GO, add collider sphere, use OnMouseDown
    • Create GO, add collider sphere, use IPointerClickHandler
      • Needs Physics raycaster on camera.
        • This is already setup for other things like clicking on planets
    • Explicit raycasting in ECS
      • Custom BaseRaycaster implementation
        • Needs to output a GameObject as the hit target, not a very good fit
  • Added components to visual track necessary for syncing GameObject position
  • Updated to Unity 6.1
  • Fixed bug in TextElementPool, reusing items that were still in use
  • Added tooltips when hovering over tactical display icon

Friday 28th​

  • Added support for dash ratio to polyline shader
  • Experimented with generating polyline paths backwards, to stabilise shaking. Doesn't seem to help.
    • Using dashes in distance instead of time, this is much more stable
      • Setting dash length based on speed, so faster things are visually obvious even when paused
  • Improved appearance of perimeters (alternating dashes)