The band gap of the rutile and anatase, the common phases of TiO2, are approximately 3.0 eV and 3.2 eV respectively. These values are higher than the photon-energy of most of the visible light, which is varied between 1.65 eV and 3.54 eV. Thus, TiO2 is expected to be transparent, as it can NOT absorb the visible light, unless it becomes a reflector the case in which it appears white. There are many possible sources for the reflection including the pores, intergranular phases, and the low grain size (less than one micron).
Based on that, it is expected that sintering eliminated the pores and increased the grain size of your film.
For the same reason that crushed glass powder is white and a bulk window is transparent. Irregularity in particle shape means that incident light is scattered in all directions and the powder appears white.. In the bulk (fused thin film) phase the attenuation coefficient is close to zero and the material appears transparent. One also has to distinguish between primary size (if we believe this to be 18 - 20 nm as given) and the fused set of post- and sub-micron aggregates and agglomerates present in the powdered form. Incidentally a set of homogeneous glass spheres (e.g. NBS 1003 c) appears transparent as the light is not scattered in all directions.
Generally TiO2 is a large bandgap semiconductor that is transparent if properly crystallized. The particles synthesized by you is not crystallized well due to the less operated temperature. Thus, annealing at higher temperature favors the solid state reaction of particles and improved the visiblity through crystal planes