We aim to explore the variable photometric and stellar properties of four yellow hypergiants (YHGs), HR8752, HR 5171A, $\rho$ Cas, and HD 179821, and their pulsations of hundreds of days, and long-term variations (LTVs) of years. We tackled multi-colour and visual photometric data sets, looked for photometric indications betraying eruptions or enhanced mass-loss episodes, calculated stellar properties mainly using a published temperature calibration, and investigated the nature of LTVs and their influence on quasi-periods and stellar properties. The $BV$ photometry revealed a high-opacity layer in the atmospheres. When the temperature rises the mass loss increases as well, consequently, as the density of the high-opacity layer. As a result, the absorption in $B$ and $V$ grow. The absorption in $B$, presumably of the order of one to a few 0\fm1, is always higher than in $V$. This difference renders redder and variable $(B-V)$ colour indexes, but the absorption law is unknown. This property of YHGs is unpredictable and explains why spectroscopic temperatures are always higher than photometric ones. We propose shorter distances for $\rho$ Cas and HR 5171A than the accepted ones. Therefore, a correction to decrease the blue luminescence of HR 5171A by polycyclic aromatic hydrocarbon (PAH) molecules is necessary, and HR 5171A would no longer be a member of the cluster Gum48d. HR 5171A is only subject to one source of light variation, not by two as the literature suggests. Eruptive episodes of YHGs prefer relatively cool circumstances when a red evolutionary loop (RL) has shifted the star to the red on the HR diagram. After the eruption, a blue loop evolution (BL) is triggered lasting one to a few decades. The reddening episode of HR 5171A between 1960 and 1974 was most likely due to a red loop evolution, and the reddening after the 1975 eruption was likely due to a shell ejection.

We re-examined photometry (VBLUW, UBV, uvby) of the yellow hypergiant HR 5171A made a few decades ago. In that study no proper explanation could be given for the enigmatic brightness excesses in the L band (VBLUW system, lambda_eff=3838 A). In the present paper, we suggest that this might have been caused by blue luminescence (BL), an emission feature of neutral polycyclic aromatic hydrocarbon molecules (PAHs), discovered in 2004. It is a fact that the highest emission peaks of the BL lie in the L band. Our goals were to investigate other possible causes, and to derive the fluxes of the emission. We used two-colour diagrams based on atmosphere models, spectral energy distributions, and different extinctions and extinction laws, depending on the location of the supposed BL source: either in Gum48d on the background or in the envelope of HR 5171A. False L-excess sources, such as a hot companion, a nearby star, or some instrumental effect, could be excluded. Also, emission features from a hot chromosphere are not plausible. The fluxes of the L excess, recorded in the data sets of 1971, 1973, and 1977 varied (all in units of 10^(-10) W m^(-2) micron^(-1)) between 1.4 to 21, depending on the location of the source. A flux near the low side of this range is preferred. Small brightness excesses in uv (uvby system) were present in 1979, but its connection with BL is doubtful. For the L fluxes we consider the lowest values as more realistic. The uncertainties are 20-30 %. Similar to other yellow hypergiants, HR 5171A showed powerful brightness outbursts, particularly in the 1970s. A release of stored H-ionization energy by atmospheric instabilities could create BL emitted by neutral PAHs.

We initiated long-term optical interferometry monitoring of the diameters of unstable yellow hypergiants (YHG) with the goal of detecting both the long-term evolution of their radius and shorter term formation related to large mass-loss events. We observed HR5171 A with AMBER/VLTI. We also examined archival photometric data in the visual and near-IR spanning more than 60 years, as well as sparse spectroscopic data. HR5171A exhibits a complex appearance. Our AMBER data reveal a surprisingly large star for a YHG R*=1315+/-260Rsun\ (~6.1AU) at the distance of 3.6+/-0.5kpc. The source is surrounded by an extended nebulosity, and these data also show a large level of asymmetry in the brightness distribution of the system, which we attribute to a newly discovered companion star located in front of the primary star. The companion's signature is also detected in the visual photometry, which indicates an orbital period of Porb=1304+/-6d. Modeling the light curve with the NIGHTFALL program provides clear evidence that the system is a contact or possibly over-contact eclipsing binary. A total current system mass of 39^{+40}_{-22} solar mass and a high mass ratio q>10 is inferred for the system. The low-mass companion of HR5171 A is very close to the primary star that is embedded within its dense wind. Tight constraints on the inclination and vsini of the primary are lacking, which prevents us from determining its influence precisely on the mass-loss phenomenon, but the system is probably experiencing a wind Roche-Lobe overflow. Depending on the amount of angular momentum that can be transferred to the stellar envelope, HR5171 A may become a fast-rotating B[e]/Luminous Blue Variable (LBV)/Wolf-Rayet star. In any case, HR5171 A highlights the possible importance of binaries for interpreting the unstable YHGs and for massive star evolution in general.

CCD UBVRi photometry of the final helium flash object V4334 Sgr (Sakurai's Object), carried out during 1997 - 1999, is presented, and the light curve from its pre-discovery rise to the dust obscuration phase is constructed. The optical light curve can be divided into four sections, the rise to maximum, the maximum, the dust onset, and the massive dust shell phase. The color indices show a general increase with time, first because of the photospheric expansion and cooling, and later because of the dust forming events. The energy distributions for the years 1996 - 1999 show that an increasing part of the energy is radiated at infrared wavelengths. In 1996, the infrared excess is likely caused by free-free radiation in the stellar wind. Starting from 1997 or 1998 at the latest, carbon dust grains are responsible for the more and more dramatic decrease of optical radiation and the growing infrared excess. Its photometric behavior in 1998 - 1999 mimics the ``red declines'' of R CrB variables, the amplitude, however, is more extreme than any fading ever observed in an R CrB star. Evidence is given that a complete dust shell has formed around V4334 Sgr. It therefore shows similarities with dust-forming classical novae, although evolving ~20 times more slowly. Its luminosity increased by a factor 4 between 1996 and 1998. A comparison of time scales of the final helium flash objects FG Sge, V605 Aql and V4334 Sgr shows that the observed photometric and spectroscopic features are similar, while V4334 Sgr is the most rapidly evolving object to date.