I agree with previous post,<div><br></div><div>usually, uint8 makes sense because it is the &quot;usual image format&quot;, uint32 allows to handle image of labels (where there might be more than 255 labels), I don&#39;t really see an use for uint16 from what I&#39;ve experienced (we might convert them to uint32, I&#39;m not sure the x2 memory loss might be too problematic here), and float32 is often used. I more rarely use float64, except sometime inside an algorithm (rather than in input/output images).</div>

<div><br></div><div>For some algorithms, it makes sense to have a uint8 version, for some it doesn&#39;t. I&#39;d say we should let the user make the conversions from/to the algorithm intended format himself, so he knows that the algorithm isn&#39;t intended for its  original  data format, and he  takes  special care in understanding why (rather than using an  inapropriate  algorithm and not worrying  about  the possible effects)</div>

<div><br></div><div>but we should clearly have conversion routines, that might also ensure that arrays are contiguous (to speed-up C/ Cython ), and possibly to ensure that we copy the array so we can modify it</div>