Expected image output of `scripts/delay_and_sum.py`?
Thanks for putting together these nice implementations and learning resources! I am interested in using DAS, FDMAS, and p-DAS to see how they affect our image quality. I get unexpectedly different image reconstructions depending on the exact code I run. Could you help diagnose / debug why, and which image is expected?
Steps to Reproduce
Run:
- https://gitlab.com/pecarlat/ultraspy/-/blob/6da936d2/scripts/delay_and_sum.py
- vs
reader.probe.show_delays(reader.acquisition_info['delays'])
before beamforming https://gitlab.com/charlesbmi/ultraspy/-/blob/example/delay-and-sum/scripts/delay_and_sum.py?ref_type=heads#L33
Expected Outcome
Output the same image in both cases, something like: https://ultraspy.readthedocs.io/en/latest/_images/interpolation_beamforming.png
Actual Outcome
Running https://gitlab.com/pecarlat/ultraspy/-/blob/6da936d2/scripts/delay_and_sum.py
Running reader.probe.show_delays(reader.acquisition_info['delays'])
before beamforming for some-reason changes the image
https://gitlab.com/charlesbmi/ultraspy/-/blob/example/delay-and-sum/scripts/delay_and_sum.py?ref_type=heads#L33
My setup
Setup in which you experienced the bug: Tried both:
- Ubuntu 22.04.3 LTS, NVIDIA GeForce RTX 4090, NVIDIA-SMI 555.42.02, CUDA Version: 12.5, Python 3.9.19, cupy-cuda12x-13.1.0
- macOS Apple Silicon CPU (required some minor code-changes to replace the
gpu_utils.py
references