
Fechado
Publicado
Pago na entrega
I have an iOS fractal explorer written in Swift that uses Metal for GPU rendering. It already implements a deep-zoom Mandelbrot path based on perturbation and float-float style precision splitting (reference orbit + shader-side deltas). At moderate zoom levels the app behaves well, but beyond roughly 1e5 magnification the two Mandelbrot paths begin to fail in different ways: - The normal Mandelbrot renderer becomes visibly pixelated and resolution-limited. - The deep Mandelbrot renderer based on perturbation does not merely become pixelated; instead, it starts to lose fine structural detail. Regions that should contain rich micro-structure turn into flat, smoothed, or incorrect-looking areas, sometimes with visible tile artefacts. During continuous zooming and panning, performance also drops sharply. So the problem is not just display resolution. The deeper issue appears to be that the perturbation-based renderer loses effective information depth / numerical fidelity at higher magnifications, possibly combined with inefficiencies in the current Swift/Metal pipeline (dispatch, buffering, tile scheduling, or memory access patterns). A core requirement is a seamless transition from the normal Mandelbrot renderer to the perturbation-based deep renderer at roughly 1e4 magnification. The handoff should be visually continuous, without a noticeable rendering discontinuity, and should allow continued deep zooming while preserving fine detail and interactive performance. I’m looking for someone who can analyze the current Swift/Metal pipeline, identify the real bottleneck(s), and implement a robust fix so that deep Mandelbrot zooming remains visually correct and interactive over much larger magnification ranges. You are free to improve or replace the current precision strategy if needed — for example with double-double arithmetic, a more stable perturbation approach, rebasing / rescaling techniques, or a CPU-side high-precision reference fallback — as long as the user experience stays responsive on device. Deliverables • Refactored Swift/Metal code for the Mandelbrot deep-zoom path • A seamless handoff from the normal renderer to the perturbation-based renderer around 1e4 magnification • Noticeably improved visual stability and retained fine detail at high magnification • Improved interactive performance during zooming and panning • A short technical note explaining the chosen precision strategy, performance tradeoffs, and possible future extensions Preferred outcomes – The standard Mandelbrot path may remain limited by screen-space resolution, but should hand off cleanly to the deep renderer at around 1e4 magnification – The deep Mandelbrot path should preserve fine structure well beyond the current limit – No visible corruption, flat detail collapse, or unstable tile artefacts during normal interactive use – Continued zooming should remain visually smooth and comfortably interactive on recent iPhones / iPads – Xcode project builds cleanly If you are experienced with floating-point numerics, perturbation rendering, Swift, Metal, and GPU optimization on Apple mobile (iOS / iPadOS) devices, I’d be very interested in your approach.
ID do Projeto: 40300745
6 propostas
Projeto remoto
Ativo há 27 dias
Defina seu orçamento e seu prazo
Seja pago pelo seu trabalho
Descreva sua proposta
É grátis para se inscrever e fazer ofertas em trabalhos
6 freelancers estão ofertando em média $42 USD for esse trabalho

Hello, I would be honored to help optimize your deep‑zoom Mandelbrot renderer. Your project aligns perfectly with my experience in floating‑point numerics, perturbation rendering, Swift, and Metal. I can diagnose where fidelity is collapsing beyond 1e5 magnification, reinforce the precision pipeline, and ensure seamless continuity at the 1e4 threshold. My approach includes analyzing GPU dispatch patterns, tile scheduling, and precision drift in the reference‑orbit perturbation path, then implementing a more stable arithmetic strategy while keeping interaction smooth on‑device. I’ll refactor the Swift/Metal pipeline, stabilize deep‑zoom structure retention, and maintain responsive panning and zooming. Kind regards, Ahtesham.
$29 USD em 10 dias
6,3
6,3

Hi, I would like to grab this opportunity and will work till you get 100% satisfied with my work. I just applied after read your job posting carefully and I believe that I am good fit to your project. I'm a serious bidder. I will satisfy you with my high skills! I am an expert which have 8+ years of experience on iPhone, Mathematics, Debugging, Swift, iOS Development, Graphics Programming, Deep Learning, Performance Tuning, Metal, Xcode I am looking forward to meet you to discuss the further detail about this project. Looking forward to hearing from you. Thank You
$25 USD em 7 dias
0,0
0,0

The loss of micro-structure and tile artifacts at 1e5 magnification points directly to precision degradation in your shader-side deltas and a lack of dynamic reference rebasing. To restore infinite depth and smooth performance, my approach includes: Implementing dynamic orbit rebasing to reset the reference point before floating-point truncation destroys the delta calculations. Integrating simulated double-double arithmetic in your Metal kernels for a mathematically flawless handoff at 1e4 magnification. Optimizing Swift-to-Metal buffer dispatch and tile scheduling to eliminate panning lag. I will deliver a refactored, visually stable pipeline quickly, ensuring the deep-zoom feels natively responsive on all modern iOS devices. Should we discuss whether a purely GPU-based approach or a CPU fallback suits your target hardware best? MS
$30 USD em 7 dias
0,0
0,0

München, Germany
Método de pagamento verificado
Membro desde mar. 15, 2026
$10-30 USD
₹400-750 INR / hora
₹1500-12500 INR
$750-1500 USD
₹12500-37500 INR
$10-30 USD
₹400-750 INR / hora
$15-25 USD / hora
$25-50 USD / hora
₹1500-12500 INR
$15-25 USD / hora
₹37500-75000 INR
$10-30 CAD
$15-25 USD / hora
€12-18 EUR / hora
₹37500-75000 INR
$15-25 USD / hora
₹12500-37500 INR
$1500-3000 USD
$15-25 USD / hora