I am worried here. Especially with the pixelated one. A program like Virtual dub takes your image sequence and just assemble it. If you have some sort of pixelation is not because of the program but of the settings you are giving.
Prepare the files.
Let's start from step 1.
I. Not every program will read a raw file, first convert it to something more useful. JPG to be specific.
Irfanview does a pretty good job batch "converting" RAW images from Nikon and Canon into JPG. I ignore if the Sony ones work as well.
II. Resample all images into a specific dimension before stitching them into a video. Crop and resample the images to the desired output. Video files some times need some specific dimensions.
You can even sharpen them a bit, and rename them to a numbered sequence... do not add too much compression on the JPG files, be careful there.
III. Now use these new files as source files for the program to stitch into a video.
IV. Now take VirtualDub2 and pull the image sequence, assign the correct frame rate and codec and export a video. Export as mp4 h264. Done. VirtualDub has not been updated since ages, but VirtualDub2 has.
If it still does not work as desired... (I do not see how), take Davinci Resolve, which needs a bit more resources, on the disk and on the system, but it is a free state of the art program and just pull the first frame of the image sequence.
V. If you need stabilization, prepare the images with more resolution than needed and then you can even stabilize it on Davinci Resolve.
Reading this sometime later I am wondering if the images have a nonstandard size for a particular video format. Some video formats need the video to be, let's say a multiple of 4.