Ekos is currently previewing, autofocusing, and auto-guiding with the image resolution specified by the driver.
Therefore, it takes a very long time to perform these processes using SBC.
I think that the processing speed will be dramatically improved by changing this to the flow to extract and process images from the stream.
(It may be converted to Fits for convenience for post-processing.)
Regarding the live stacking that MR.Han is currently working on, I think that if you can add a function to extract images from the stream, you can enjoy it at a realistic speed.
The images (videos) used for display and analysis are lightweight and fast, and for shooting, you can perform comfortable operations by saving the file size set by the driver and changing the role for each application.
I think the INDI driver operates very fast compared to ASCOM.
However, since many users use the still image + loop of the maximum pixel of the camera for autofocus and autoguide, they tend to recognize it as slow.
Since the use with native drivers such as SharpCap and PHD2 is close to the flow of this proposal, it is previewed at high speed.
The flow is an astronomical camera and there is no internal processing like DSLR.
So I think it will be necessary to devise a software that can be handled comfortably by software processing such as Ekos.
Live view (INDI live video) displays a low-resolution stream image (for example, 800x600), and when zoomed, it automatically switches to a high-resolution setting and displays the ROI image with the above number of pixels (800x600).
In this way, the network driver can reduce the amount of live view transfer.
It can also extract files from lightweight streams and process them very quickly over the network.
This flow is suitable for solver analysis, autofocus analysis, and autoguide.
ASIAir already uses this concept, so solver processing and live view are very fast.